CN107818289A

CN107818289A - A kind of prescription recognition methods and device, a kind of device for prescription identification

Info

Publication number: CN107818289A
Application number: CN201610823024.6A
Authority: CN
Inventors: 银磊; 李明修; 卜海亮
Original assignee: Beijing Sogou Technology Development Co Ltd
Current assignee: Beijing Sogou Technology Development Co Ltd
Priority date: 2016-09-13
Filing date: 2016-09-13
Publication date: 2018-03-20

Abstract

The embodiments of the invention provide a kind of prescription recognition methods and device, a kind of device for prescription identification, method therein specifically includes：Cutting is carried out to the image of prescription to be identified according to character, obtains the character picture corresponding to each character；Each character picture is identified respectively using medical character recognition storehouse, obtains alternative characters corresponding to each character picture；The medical character recognition storehouse is got according to the image pattern training of medical character；Alternative characters corresponding to all character pictures that image according to the prescription to be identified is included, obtain prescription recognition result corresponding to the image of the prescription to be identified.The embodiment of the present invention can improve the accuracy rate of medical character recognition and prescription identification.

Description

A kind of prescription recognition methods and device, a kind of device for prescription identification

Technical field

The present invention relates to pharmaceutical technology field, is used for prescription more particularly to a kind of prescription recognition methods and device, one kind The device of identification.

Background technology

The traditional Chinese medical science is the traditional medicine in China, the experience accumulated over the past thousands of years according to people when treating disease, forms crowd More basic prescriptions for being used to treat corresponding disease, these basic prescriptions are reference of the traditional Chinese medical science when treating disease, and the traditional Chinese medical science is at this On the basis of a little basic prescriptions, corresponding adjustment is made to dose and medicine according to the actual symptoms of patient, to obtain being adapted to trouble The traditional Chinese medical science prescription of person.

Current traditional Chinese medical science prescription is by hand-written completion mostly.Existing scheme can turn the traditional Chinese medical science prescription of hand-written completion Picture or photo are changed to, and utilizes OCR (optics medicine character recognition, Optical Character Recognition) skill Art, by the word content on picture or photo, editable text is converted to, so as to complete the identification of traditional Chinese medical science prescription.

But because everyone hand-written script has the characteristics of respective, particularly rapid style of writing hand-written script not only to connect a book Write, most of stroke gross distortion, and there are many strokes to be removed.The characteristics of above-mentioned hand-written script, causes existing scheme very Difficult to identify corresponding medicine name in traditional Chinese medical science prescription exactly, this causes the recognition accuracy of traditional Chinese medical science prescription relatively low.

The content of the invention

In view of the above problems, it is proposed that the embodiment of the present invention overcomes above mentioned problem or at least in part to provide one kind Prescription recognition methods, prescription identification device and the device for prescription identification to solve the above problems, the embodiment of the present invention can Improve the accuracy rate of medical character recognition and prescription identification.

In order to solve the above problems, the invention discloses a kind of prescription recognition methods, including：

Cutting is carried out to the image of prescription to be identified according to character, obtains the character picture corresponding to each character；

Each character picture is identified respectively using medical character recognition storehouse, obtained standby corresponding to each character picture Word selection accords with；The medical character recognition storehouse is got according to the image pattern training of medical character；

Alternative characters corresponding to all character pictures that image according to the prescription to be identified is included, obtain described treat Identify prescription recognition result corresponding to the image of prescription.

Alternatively, when each character picture corresponds to multiple alternative characters, the image according to the prescription to be identified Comprising all character pictures corresponding to alternative characters, obtain prescription recognition result corresponding to the image of the prescription to be identified The step of, including：

The attribute of alternative characters corresponding to each character picture included in image according to the prescription to be identified is special Multiple alternative characters corresponding to each character picture are screened, to obtain target alternative corresponding to each character picture by sign Character；

Target alternative character corresponding to all character pictures difference included in image according to the prescription to be identified, is obtained To prescription recognition result corresponding to the image of the prescription to be identified；

Wherein, the attributive character includes at least one of following feature：Identification probability corresponding to the alternative characters, Medical vocabulary characteristic, medicine property of medicine, medicine similarity and medicine co-occurrence probability with preset prescription.

Alternatively, the attributive character includes：Medical vocabulary characteristic corresponding to the alternative characters, then it is described to each word The step of multiple alternative characters corresponding to symbol image are screened, including：

For multiple alternative characters corresponding to each character picture, each character picture is determined using preset medical dictionary Corresponding target alternative character.

Alternatively, it is described to determine the step of target alternative character corresponding to each character picture using preset medical dictionary Suddenly, including：

It is for each character picture, multiple alternative characters corresponding to the character picture are corresponding with adjacent character image Multiple alternative characters, form multiple combining characters strings；

For each combining characters string, judging to whether there is in the preset medical dictionary includes the combining characters string Medical vocabulary；If it is present using alternative characters corresponding to the combining characters string for hitting preset medical dictionary as the character Target alternative character corresponding to image.

For each combining characters string, judging to whether there is in the preset medical dictionary includes the combining characters string Medical vocabulary；If it is present the identification probability of alternative characters corresponding to the combining characters string is weighted；

All combining characters strings are judged after terminating, select the maximum alternative characters of identification probability as the character picture Corresponding target alternative character.

Alternatively, the attributive character includes：Medicine property of medicine corresponding to the alternative characters, then it is described to each character figure The step of multiple alternative characters are screened as corresponding to, including：

Image according to prescription to be identified includes the property of medicine of medicine corresponding to character picture, determines character picture to be screened Each drug candidate and other character pictures corresponding to property of medicine diversity factor between medicine, wherein, corresponding to the character picture Medicine is that the combination of alternative characters corresponding to foundation adjacent character image obtains；

According to the property of medicine diversity factor, target alternative word is chosen from multiple alternative characters corresponding to character picture to be screened Symbol.

Alternatively, the attributive character includes：Corresponding to the alternative characters with the medicine similarity of preset prescription, then institute State to the step of multiple alternative characters are screened corresponding to each character picture, including：

Determine that alternative characters correspond to the medicine similarity between prescription to be identified and preset prescription；

When the similarity is more than similarity threshold, institute is chosen from multiple alternative characters corresponding to the character picture State preset prescription possessed target alternative character.

Alternatively, the attributive character includes：Medicine co-occurrence probability corresponding to the alternative characters, it is described to each character The step of multiple alternative characters are screened corresponding to image, including：

Determine character picture to be screened each drug candidate and other character pictures corresponding to medicine co-occurrence between medicine Probability；

According to the medicine co-occurrence probability, target alternative word is chosen from multiple alternative characters corresponding to the character picture Symbol.

Alternatively, it is described to the step of multiple alternative characters are screened corresponding to each character picture, including：

In medicine similarity and medicine co-occurrence probability according to the medical vocabulary characteristic, medicine property of medicine and preset prescription At least one, identification probability corresponding to corresponding alternative characters is adjusted；

According to the identification probability after adjustment, target alternative word is chosen from multiple alternative characters corresponding to the character picture Symbol.

Alternatively, the image of the prescription to be identified includes：The image of the prescription to be identified of multiple shooting angle, the side Method also includes：

The quantity of medicine is included according to prescription recognition result, from the image of the prescription to be identified of the multiple shooting angle Optimal prescription recognition result is chosen in corresponding multiple prescription recognition results.

Alternatively, the prescription recognition result includes：The title of at least one medicine, methods described also include：

According to the mapping relations corresponding to medicine referred to as between full name, by the title of medicine in the prescription recognition result Abbreviation be converted to full name.

Alternatively, it is described that cutting is carried out to the image of prescription to be identified according to character, obtain the word corresponding to each character Image is accorded with, including：

Cutting is carried out to the image of prescription to be identified according to separator, obtains the medicine character figure corresponding to every kind of medicine Picture；

In units of character, cutting is carried out to the medicine character picture, obtains monocase image corresponding to each character.

Alternatively, alternative word corresponding to all character pictures that the image according to the prescription to be identified is included Symbol, obtains prescription recognition result corresponding to the image of the prescription to be identified, including：

For each medicine character picture, the alternative characters for all monocase images that it is included are combined, with Combining characters string corresponding to composition；

When the medicine character picture corresponds to multiple combining characters strings, for each combining characters string, preset doctor is judged With the presence or absence of the medical vocabulary for including the combining characters string in medicine dictionary；

It is preset to hitting when multiple combining characters strings hit preset medical dictionary corresponding to the medicine character picture Multiple combining characters strings of medical dictionary are screened, to obtain corresponding objective cross character string；

By combined characters corresponding to objective cross character string corresponding to the medicine character picture and other drugs character picture Symbol string is combined, to obtain corresponding prescription recognition result.

Alternatively, the step of multiple combining characters strings of the described pair of preset medical dictionary of hit screen, including：

When multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, according to each combining characters Property of medicine diversity factor between medicine corresponding to the corresponding drug candidate of string and other drugs character picture, from the multiple combining characters Objective cross character string is chosen in string；And/or

When multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, each combining characters are determined String drug candidate and the medicine similarity corresponding to medicine between prescription to be identified and preset prescription is determined；It is big in the similarity When similarity threshold, the preset prescription possessed objective cross character string is chosen from the multiple combining characters string； And/or

When multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, each combining characters are determined Medicine co-occurrence probability between medicine corresponding to the corresponding drug candidate of string and other character pictures；It is general according to the medicine co-occurrence Rate, objective cross character string is chosen from the multiple combining characters string.

On the other hand, the invention discloses a kind of prescription identification device, including：

Image cutting module, for carrying out cutting to the image of prescription to be identified according to character, it is right to obtain each character institute The character picture answered；

Picture recognition module, for each character picture to be identified respectively using medical character recognition storehouse, obtain every Alternative characters corresponding to individual character picture；The medical character recognition storehouse is got according to the image pattern training of medical character； And

As a result acquisition module, for standby corresponding to all character pictures for being included according to the image of the prescription to be identified Word selection accords with, and obtains prescription recognition result corresponding to the image of the prescription to be identified.

Alternatively, when each character picture corresponds to multiple alternative characters, the result acquisition module includes：

Alternative characters screen submodule, for each character picture included in the image according to the prescription to be identified The attributive character of corresponding alternative characters, multiple alternative characters corresponding to each character picture are screened, it is each to obtain Target alternative character corresponding to character picture；

As a result acquisition submodule, all character pictures for being included in the image according to the prescription to be identified are right respectively The target alternative character answered, obtain prescription recognition result corresponding to the image of the prescription to be identified；

Alternatively, the attributive character specifically includes：Medical vocabulary characteristic corresponding to the alternative characters, then it is described alternative Character screening submodule includes：

Determining unit, for for multiple alternative characters corresponding to each character picture, using preset medical dictionary come really Target alternative character corresponding to fixed each character picture.

Alternatively, the determining unit includes：

Subelement is combined, for for each character picture, by multiple alternative characters corresponding to the character picture and phase Multiple alternative characters corresponding to adjacent character picture, form multiple combining characters strings；

First judgment sub-unit, for for each combining characters string, judging to whether there is in the preset medical dictionary Include the medical vocabulary of the combining characters string；If it is present it will hit corresponding to the combining characters string of preset medical dictionary Alternative characters are as target alternative character corresponding to the character picture.

Alternatively, the determining unit includes：

Second judgment sub-unit, for for each combining characters string, judging to whether there is in the preset medical dictionary Include the medical vocabulary of the combining characters string；If it is present the identification to alternative characters corresponding to the combining characters string Probability is weighted；

Subelement is selected, for judging after terminating all combining characters strings, selects the maximum alternative characters of identification probability As target alternative character corresponding to the character picture.

Alternatively, the attributive character includes：Medicine property of medicine corresponding to the alternative characters, then the alternative characters screening Submodule includes：

Diversity factor determining unit, for including the medicine of medicine corresponding to character picture according to the image of prescription to be identified Property, determine character picture to be screened each drug candidate and other character pictures corresponding to property of medicine diversity factor between medicine, its In, medicine corresponding to the character picture is that the combination of alternative characters corresponding to foundation adjacent character image obtains；

First chooses unit, for according to the property of medicine diversity factor, from multiple alternative words corresponding to character picture to be screened Target alternative character is chosen in symbol.

Alternatively, the attributive character includes：Corresponding to the alternative characters with the medicine similarity of preset prescription, then institute Stating alternative characters screening submodule includes：

Similarity determining unit, for determining that the medicine that alternative characters are corresponded between prescription to be identified and preset prescription is similar Degree；

Second chooses unit, for when the similarity is more than similarity threshold, from more corresponding to the character picture The preset prescription possessed target alternative character is chosen in individual alternative characters.

Alternatively, the attributive character includes：Medicine co-occurrence probability corresponding to the alternative characters, then alternative characters Screening submodule includes：

Co-occurrence probability determining unit, each drug candidate and other character pictures institute for determining character picture to be screened are right Answer the medicine co-occurrence probability between medicine；

3rd chooses unit, for according to the medicine co-occurrence probability, from multiple alternative words corresponding to the character picture Target alternative character is chosen in symbol.

Alternatively, the alternative characters screening submodule includes：

Probability adjustment unit, for according to the medical vocabulary characteristic, medicine property of medicine, the medicine similarity with preset prescription At least one of with medicine co-occurrence probability, identification probability corresponding to corresponding alternative characters is adjusted；

4th chooses unit, the identification probability after being adjusted for foundation, from multiple alternative words corresponding to the character picture Target alternative character is chosen in symbol.

Alternatively, the image of the prescription to be identified includes：The image of the prescription to be identified of multiple shooting angle, then it is described Device also includes：

As a result module is chosen, for including the quantity of medicine according to prescription recognition result, from the multiple shooting angle Prescription to be identified image corresponding to choose optimal prescription recognition result in multiple prescription recognition results.

Alternatively, the prescription recognition result includes：The title of at least one medicine, then described device also include：

Modular converter, for according to the mapping relations corresponding to medicine referred to as between full name, the prescription being identified and tied The abbreviation of the title of medicine is converted to full name in fruit.

Alternatively, described image cutting module includes：

First cutting submodule, for carrying out cutting to the image of prescription to be identified according to separator, obtain every kind of medicine Corresponding medicine character picture；

Second cutting submodule, in units of character, carrying out cutting to the medicine character picture, obtaining each word Monocase image corresponding to symbol.

Alternatively, the result acquisition module includes：

First combination submodule, for for each medicine character picture, all monocase images that it is included Alternative characters are combined, with combining characters string corresponding to composition；

Vocabulary judging submodule, for when the medicine character picture corresponds to multiple combining characters strings, for each group Character string is closed, is judged in preset medical dictionary with the presence or absence of the medical vocabulary for including the combining characters string；

Combining characters string screens submodule, for being hit when multiple combining characters strings corresponding to the medicine character picture During preset medical dictionary, the multiple combining characters strings for hitting preset medical dictionary are screened, to obtain corresponding target group Close character string；

Second combination submodule, for by objective cross character string corresponding to the medicine character picture and other drugs word Combining characters string corresponding to symbol image is combined, to obtain corresponding prescription recognition result.

Alternatively, the combining characters string screening submodule includes：

First combining characters string screening unit, for when multiple combining characters strings corresponding to medicine character picture hit it is pre- When putting medical dictionary, the medicine between medicine corresponding to drug candidate and other drugs character picture is corresponded to according to each combination character string Sex differernce degree, objective cross character string is chosen from the multiple combining characters string；And/or

Second combining characters string screening unit, for when multiple combining characters strings corresponding to medicine character picture hit it is pre- When putting medical dictionary, determine each combination character string drug candidate and determined corresponding to medicine between prescription to be identified and preset prescription Medicine similarity；When the similarity is more than similarity threshold, chosen from the multiple combining characters string described preset Prescription possessed objective cross character string；And/or

3rd combining characters string screening unit, for when multiple combining characters strings corresponding to medicine character picture hit it is pre- When putting medical dictionary, determine that the medicine that each combination character string is corresponded between medicine corresponding to drug candidate and other character pictures is same Existing probability；According to the medicine co-occurrence probability, objective cross character string is chosen from the multiple combining characters string.

Another further aspect, the invention discloses a kind of device for prescription identification, include memory, and one or More than one program, one of them or more than one program storage are configured to by one or one in memory Individual above computing device is one or more than one program bag contains the instruction for being used for being operated below：

The embodiment of the present invention includes advantages below：

The embodiment of the present invention can carry out medicine to be identified using special medical character recognition storehouse in units of medical character The identification of the image of side, due to involved in field of medicaments medical character it is limited (by taking the identification of traditional Chinese medical science prescription as an example, traditional Chinese medicine The title of material can be enumerated), identify that the difficulty of the character picture corresponding to single character entirely is waited to know well below identification The difficulty of the image of other prescription, also, the medical character recognition storehouse that can also be directed to medical character optimizes, such as can make It is trained with a large amount of script samples of preset character, therefore the standard of medical character recognition and prescription identification can be effectively improved True rate.

In addition, in the case where alternative characters corresponding to character picture are more than one, can be according to corresponding to alternative characters Identification probability, medical vocabulary characteristic, medicine property of medicine, with the attributive character such as the medicine similarity of preset prescription and medicine co-occurrence probability The possibility of alternative characters is determined, and chooses the maximum alternative characters of possibility as target alternative character, it is possible thereby to improve The accuracy rate of medical character recognition.

Brief description of the drawings

Fig. 1 is a kind of step flow chart of prescription recognition methods embodiment one of the present invention；

Fig. 2 is a kind of step flow chart of prescription recognition methods embodiment two of the present invention；

Fig. 3 is a kind of step flow chart of prescription recognition methods embodiment three of the present invention；

Fig. 4 is a kind of step flow chart of prescription recognition methods example IV of the present invention；

Fig. 5 is a kind of structured flowchart of prescription identification device embodiment of the present invention；

Fig. 6 is a kind of block diagram of device 900 for prescription identification of the present invention；And

Fig. 7 is the structural representation of server in some embodiments of the present invention.

Embodiment

In order to facilitate the understanding of the purposes, features and advantages of the present invention, it is below in conjunction with the accompanying drawings and specific real Applying mode, the present invention is further detailed explanation.

Embodiment of the method one

Reference picture 1, a kind of step flow chart of prescription recognition methods embodiment one of the present invention is shown, can specifically be wrapped Include following steps：

Step 101, according to character cutting is carried out to the image of prescription to be identified, obtain the character figure corresponding to each character Picture；

Step 102, using medical character recognition storehouse each character picture is identified respectively, obtains each character picture Corresponding alternative characters；The medical character recognition storehouse is got according to the image pattern training of medical character；

Alternative characters corresponding to all character pictures that step 103, the image according to the prescription to be identified are included, are obtained To prescription recognition result corresponding to the image of the prescription to be identified.

The embodiments of the invention provide a kind of prescription identifying schemes, the program can utilize special medical character recognition storehouse The medical name included in the image of prescription to be identified is identified in units of single medical character, by institute in field of medicaments The medical character being related to is limited (by taking the identification of traditional Chinese medical science prescription as an example, the title of traditional Chinese medical science medicinal material can be enumerated), and identification is single Image of the difficulty of character picture corresponding to character (hereinafter referred to as monocase image) well below the whole prescription to be identified of identification Difficulty, also, the medical character recognition storehouse that can also be directed to medical character optimize, such as can use single medical word A large amount of script samples of symbol are trained, therefore can effectively improve the accuracy rate of medical character recognition and prescription identification.

Prescription identifying schemes provided in an embodiment of the present invention can apply to medical related web site, medicine related APP (applications Program, Application) etc. have prescription identification function application program in, and prescription provided in an embodiment of the present invention identify Scheme can be applied in client application environment corresponding with server, wherein, client and server can be located at it is wired or In wireless network, by the wired or wireless network, client carries out data interaction with server.

Specifically, client may operate on intelligent terminal, and above-mentioned intelligent terminal can specifically include but is not limited to：Intelligence Can mobile phone, tablet personal computer, E-book reader, MP3 (dynamic image expert's compression standard audio aspect 3, Moving Picture Experts Group Audio Layer III) player, MP4 (dynamic image expert's compression standard audio aspects 4, Moving Picture Experts Group Audio Layer IV) player, pocket computer on knee, vehicle-mounted computer, desk-top meter Calculation machine, set top box, intelligent TV set, wearable device etc..

In actual applications, client can pass through the image for the prescription to be identified that preset interface captures user inputs.Example Such as, when receiving the trigger action for the preset interface, it can provide and upload option or option of taking pictures, wherein, user The image of the prescription to be identified shot in advance can be uploaded according to the upload option, or, user will by the option of taking pictures The prescription of hand-written completion is converted to the image of prescription to be identified, and the image of satisfied prescription to be identified is uploaded.It is appreciated that The embodiment of the present invention is not any limitation as the specific acquisition modes of the image of prescription to be identified.

In addition, the prescription to be identified of the embodiment of the present invention can specifically include：Traditional Chinese medical science prescription, doctor trained in Western medicine prescription, traditional Chinese and western medicine knot Prescription etc. is closed, character corresponding to various medicines can be included in above-mentioned prescription.Wherein, the medicine can specifically include：Medicinal material, Medicine etc..Wherein, medicinal material is to be available for the raw material of pharmacy, and alternatively above-mentioned medicinal material can include Chinese medicine, i.e., it is not processed or The Chinese medicine material of finished product is not made, certainly, Western medicine material is also within the covering scope of the medicinal material of the embodiment of the present invention.Medicine refers to The finished product of medicine, specifically, medicine refer to the disease for preventing, treating, diagnose people, purposefully adjust the physiological function of people simultaneously Regulation has the material of indication or major function, usage and dosage, specifically includes but the unlimited prepared slices of Chinese crude drugs, Chinese patent drug, chemistry Bulk drug and its preparation, antibiotic, biochemical drug, radioactive drugs, serum, vaccine, blood product and diagnostic aid etc..For side Just describe, character corresponding to medicine is referred to as medical character by the embodiment of the present invention.

In actual applications, substantial amounts of character may be included in the image of prescription to be identified.Step 101 can be with single character For unit, the image of prescription to be identified is divided into several monocase images, can include individual character in each monocase image Symbol.Above-mentioned segmentation can divide according to monocase to the character picture in prescription image.Wherein, can in cutting procedure To be carried out according to preset order (such as order from front to back or from back to front) to the character in the image of prescription to be identified time Go through, to identify the edge of monocase, and then corresponding monocase image is obtained according to the edge.Certainly, it is above-mentioned based on edge Image partitioning scheme is intended only as alternative embodiment, in fact, the segmentation side based on threshold value, region, graph theory or energy functional Formula can realize the segmentation of the embodiment of the present invention, and the embodiment of the present invention is not any limitation as to specific partitioning scheme.

In the embodiment of the present invention, medical character recognition storehouse used in step 102 can be to possess single medical character to correspond to The learning model of the recognition capability of monocase image, it can receive the monocase image of input, export the monocase image pair The character answered.Also, above-mentioned medical character recognition storehouse can be got according to the image pattern training of single medical character, therefore can To be optimized to medical character recognition storehouse, such as a large amount of script samples of single medical character can be used to be trained, because This can effectively improve the accuracy rate of single medical character recognition and prescription identification.

In actual applications, the medical character of above-mentioned list can derive from the preset medical text of big data.It is alternatively possible to Using classical medical book, case, ancient prescription, into the corresponding text such as side as preset medical text, and build medical text database.It is optional Ground, image corresponding to above-mentioned preset medical text can be split according to single character, to obtain single medical character pair The image answered, as image pattern collection.For example, can shoot to obtain into the image of side, and by splitting the image for the side of being extracted into The image of the middle medical character of list.

Or individual character corresponding to each medicine included by preset medical text can be obtained from above-mentioned preset medical text Symbol, namely single medical character.For example, the medical characters such as " white ", " Chinese herbaceous peony ", " when ", " returning " can be obtained from Cheng Fangzhong.Further, The image pattern collection corresponding to the hand-written script or standard letter of the medical character of each list can be obtained.Due to the identification of standard letter Difficulty is smaller, and uses hand-written script in view of traditional Chinese medical science prescription more, therefore the embodiment of the present invention is mainly with the image sample of hand-written script Illustrated exemplified by this collection and its training and identification process, the image pattern collection of standard letter and its training and identification process are mutual With reference to.In actual applications, can be by hand-written in user feedback, third-party hand-written script function or existing prescription The modes such as the multiplexing of font obtain the image pattern collection corresponding to the hand-written script of the medical character of each list, it will be understood that the present invention Embodiment is not any limitation as the specific collection mode of image pattern collection.

In a kind of alternative embodiment of the present invention, in the training process of image pattern collection, image pattern can be extracted The feature of collection, wherein, this feature can specifically include：Pixel value, boundary point position, the stroke that is both horizontally and vertically averaged pass through Number and profile point (namely position of word each point) etc., dimensionality reduction, binaryzation, sharp can be carried out to the pixel value in the training process The operation such as change.Further, in the identification process of character picture, the feature of image pattern collection and the spy of character picture can be calculated Similarity between sign, the identification probability using the similarity as character picture, and by medical character identification result and corresponding Identification probability exports.

, can be according to PCA (principal component analysis, Principal Component in a kind of alternative embodiment of the present invention Analysis), the training method such as neutral net, rarefaction representation carries out the training of image pattern collection.Wherein, PCA methods can lead to Linear transformation is crossed by the expression that image pattern set transformation is one group of each dimension linear independence, the master available for extraction image pattern collection Want characteristic component；Sparse representation method can avoid in target identification problem feature extraction and grader to caused by classification results Influence, by many experiments, as long as showing that gained coefficient solution is sparse enough, the classification based on L1/2 norms can obtain preferably Discrimination, and Riming time of algorithm is reduced, and character picture can be identified exactly.It is appreciated that above-mentioned training method is made For alternative embodiment, the embodiment of the present invention is not any limitation as specific training method.

In actual applications, step 102 can identify to obtain the character picture for character picture corresponding to each character Corresponding alternative characters.For each character picture, its corresponding alternative characters can be identification probability highest one Alternative characters or the higher multiple alternative characters of identification probability, so, step 103 can be to the figure of prescription to be identified The alternative characters as corresponding to all character pictures included are combined, to obtain corresponding prescription recognition result.Assuming that treat The character picture for identifying the image of prescription is B_j, 0≤j≤m-1, alternative characters corresponding to each character picture are B_jk, 0≤k, then Can be to B_0k+B_1k+B_2k+…+B_m-1,kIt is combined, to obtain corresponding prescription recognition result.

In a kind of alternative embodiment of the present invention, when each character picture corresponds to multiple alternative characters, step 103 institute Alternative characters corresponding to all character pictures that the image for the foundation prescription to be identified stated is included, are obtained described to be identified Prescription recognition result corresponding to the image of prescription, can specifically include：

Step A1, alternative characters corresponding to each character picture included in the image according to the prescription to be identified Attributive character, multiple alternative characters corresponding to each character picture are screened, to obtain mesh corresponding to each character picture Mark alternative characters；

Step A2, according to target alternative corresponding to all character pictures difference included in the image of the prescription to be identified Character, obtain prescription recognition result corresponding to the image of the prescription to be identified.

Wherein, the attributive character can specifically include at least one of following feature：Corresponding to the alternative characters Identification probability, medical vocabulary characteristic, medicine property of medicine, medicine similarity and medicine co-occurrence probability with preset prescription.

Above-mentioned identification probability can derive from medical character recognition storehouse；Medical vocabulary characteristic can be used for characterize alternative characters or Whether its corresponding combining characters of person are medical vocabulary；Medicine property of medicine can be used for representing can be used for representing the property of medicine, smell and Function, such as herbal nature, refer to relevant with therapeutic action performance possessed by Chinese medicine, can be summarized as four natures and five flavors of drug, channel tropism, Ascending-aescending-floating-sinking, toxicity etc.；With the medicine similarity of preset prescription can be used for representing the corresponding prescription to be identified of alternative characters with it is preset Similarity between prescription；Medicine co-occurrence probability can be used for the probability for representing that multi-medicament occurs simultaneously.

In a kind of alternative embodiment of the present invention, include in above-mentioned attributive character：It is medical corresponding to the alternative characters During vocabulary characteristic, being screened to multiple alternative characters corresponding to each character picture described in step A1 is each to obtain Target alternative character corresponding to character picture, it can specifically include：For multiple alternative characters corresponding to each character picture, profit Target alternative character corresponding to each character picture is determined with preset medical dictionary.

Specifically, described determine that target alternative character can wrap corresponding to each character picture using preset medical dictionary Include：It is for each character picture, multiple alternative characters corresponding to the character picture are corresponding with adjacent character image multiple standby Word selection accords with, and forms multiple combining characters strings；For each combining characters string, judge in the preset medical dictionary with the presence or absence of bag Include the medical vocabulary of the combining characters string；If it is present it will can hit corresponding to the combining characters string of preset medical dictionary Alternative characters are as target alternative character corresponding to the character picture.

Or described determine that target alternative character can also wrap corresponding to each character picture using preset medical dictionary Include：It is for each character picture, multiple alternative characters corresponding to the character picture are corresponding with adjacent character image multiple standby Word selection accords with, and forms multiple combining characters strings；For each combining characters string, judge in the preset medical dictionary with the presence or absence of bag The medical vocabulary of the combining characters string is included, if it is present identification that can be to alternative characters corresponding to the combining characters string is general Rate is weighted.Alternatively, if being not present, the identification probability of alternative characters corresponding to the combining characters string can be dropped Power；Finally, all combining characters strings are judged after terminating, selects the maximum alternative characters of identification probability as the character picture pair The target alternative character answered.

Wherein, the adjacent character image can be the prescription to be identified image in, the character picture correspond to character Adjacent character corresponding to character picture；In the image of the prescription to be identified, character corresponding to the adjacent character image can be with Correspond to the left or right of character positioned at the character picture, in addition the prescription to be identified image shooting angle not just When, character corresponding to the adjacent character image may be located on the character picture and correspond to either above or below of character etc., can be with Understand, the embodiment of the present invention is not any limitation as the specific orientation of adjacent character image.

In view of medical vocabulary corresponding to medicine length differ (some 2, some for 3 or other numerical value), each The quantity of adjacent character image corresponding to character picture can be 1 or 2 etc., for example, can select positioned at the same of character picture One or more adjacent character image in individual orientation or different azimuth., can be respectively by the character in the embodiment of the present application At least one alternative characters corresponding with an adjacent character image of multiple alternative characters corresponding to image, form multiple combined characters Symbol string；Can also be corresponding with multiple adjacent character images multiple alternative by multiple alternative characters corresponding to the character picture respectively Character, form multiple combining characters strings.When using multiple adjacent character images, character corresponding to the plurality of adjacent character image The same orientation that the character picture corresponds to character can be located at, the not Tongfang of character can also be corresponded to positioned at the character picture respectively Position (as being located at the left and right respectively)；It is not specifically limited in the embodiment of the present invention.

In actual applications, can be using the affiliated character picture of alternative characters to be screened as character picture to be screened.Can Selection of land, when adjacent character image is multiple, it can also be selected from multiple adjacent character images and character picture category to be screened In the adjacent character image of same medical vocabulary.The selective factor B of institute's foundation can include：Character picture to be screened and adjacent words Accord with the distance between image, wherein, apart from smaller, then explanation belong to both belong to same medical vocabulary probability it is bigger, and then The selected probability of corresponding adjacent character image is bigger.

It should be noted that in the selection for the adjacent character image for belonging to same medical vocabulary with character picture to be screened Cheng Zhong, the adjacent character image of such as separator can also be filtered according to alternative characters corresponding to adjacent character image. For ", ", ", ", "：", for the character picture corresponding to separator such as " (", ") ", because step 102 utilizes medical character Identification storehouse identifies to obtain its corresponding alternative characters (namely text), therefore adjacent character image can be entered according to its alternative characters Row filtering.It is alternatively possible to filtering rule corresponding to preset separator, as the filtering rule is：Filtering ", ", ", ", "：" etc. symbol Number etc..It is appreciated that those skilled in the art can also be according to practical application request, pair other characters unrelated with medical attribute Filtered, for example, can be according to alternative characters corresponding to adjacent character image, to the phase of such as numerical character and alphabetic character Adjacent character picture filters etc..

Preset medical dictionary can be used for storing existing medical vocabulary.In actual applications, can be by existing on network Medicinal material, prescription encyclopaedia data, traditional Chinese medicine dictionary, the approach such as TCM Literature obtain existing medical vocabulary, and to acquired Medical vocabulary verified and supplemented, the medical vocabulary by verification is stored to preset medical dictionary, it will be understood that this Inventive embodiments are not any limitation as the specific collection mode of preset medical dictionary traditional Chinese medicine vocabulary.

After each character picture on the image to prescription to be identified identifies corresponding alternative characters, this optional reality Example is applied to screen multiple alternative characters corresponding to certain character picture according to preset medical dictionary, specifically, Ke Yijian Whether combining characters string corresponding to surveying alternative characters exists in preset medical dictionary, if can be to the knowledge of alternative characters in the presence of if Other probability is weighted.Otherwise drop power can be carried out to the identification probability of alternative characters, maximum standby of final choice identification probability Word selection symbol is used as target alternative character.

In one kind application example of the present invention, it is assumed that step 102 identifies a character using medical character recognition storehouse Alternative characters include corresponding to image：" not " and " art ", it is if standby corresponding to adjacent character image positioned at the character picture left Word selection symbol be " white ", then can respectively will " not " and " art " respectively with it is " white " form combining characters string：" in vain not " and " bighead atractylodes rhizome "；Enter It one step, can be obtained by searching preset medical dictionary, the medical vocabulary for including " bighead atractylodes rhizome " in preset medical dictionary be present, not In the presence of the medical vocabulary including " in vain not ", therefore alternative characters " art " corresponding to character string " bighead atractylodes rhizome " can be directly selected and be used as target Alternative characters；It is of course also possible to the identification probability of alternative characters " art " corresponding to character string " bighead atractylodes rhizome " is weighted, by character string The identification probability drop power of alternative characters " not " corresponding to " in vain not ".

In another alternative embodiment of the present invention, include in the attributive character：Medicine corresponding to the alternative characters During the thing property of medicine, multiple alternative characters corresponding to each character picture are screened described in step A1, it can specifically include： Image according to prescription to be identified includes the property of medicine of medicine corresponding to character picture, determines each alternative of character picture to be screened Property of medicine diversity factor between medicine corresponding to medicine and other character pictures, wherein, medicine corresponding to the character picture be according to Obtained according to the combination of alternative characters corresponding to adjacent character image；According to the property of medicine diversity factor, from character picture pair to be screened Target alternative character is chosen in the multiple alternative characters answered.

Property of medicine diversity factor in above-mentioned property of medicine diversity factor namely prescription to be identified between medicine, for certain character figure to be screened As for, corresponding multiple drug candidates can be obtained according to multiple alternative characters corresponding to it, then can calculate each alternative medicine Property of medicine diversity factor between medicine corresponding to thing and other character pictures；Further, when certain property of medicine diversity factor meets preset medicine Property close to condition, then it is considered that the possibility of corresponding alternative characters is bigger, conversely, then it is considered that the alternative characters can Energy property is smaller.

A kind of acquisition process of medicine corresponding to character picture is provided herein, specifically, can be by character picture Alternative characters alternative characters corresponding with adjacent character image, combining characters string corresponding to composition, and it is directed to each combining characters String, judge in preset medical dictionary with the presence or absence of the medical vocabulary for including the combining characters string；If it is present it will can hit Mark of the combining characters string of preset medical dictionary as corresponding medicine.Assuming that the character picture of the image of prescription to be identified is B_j, 0≤j≤m-1, alternative characters corresponding to each character picture are B_jk, 0≤k then can be by B_jkIt is standby with adjacent character object Word selection symbol forms combining characters string.For example, can be by B_0kWith B_1kCombining characters string is formed, or, by B_2kWith B_3kForm combined characters Symbol string etc..

In a kind of alternative embodiment of the present invention, when multiple alternative characters and the adjacent character image of a character picture Corresponding to corresponding alternative characters composition during multiple combining characters strings, if multiple combining characters strings hit preset medical dictionary, The property of medicine diversity factor between medicine corresponding to drug candidate and other character pictures can be then corresponded to according to each combination character string, it is right Multiple alternative characters of the character picture are screened.

In one kind application example of the present invention, it is assumed that identify certain character picture in the medical character recognition storehouse of step 102 Alternative characters include：" wood " or " art ", the alternative characters positioned at the adjacent character image of the character picture left are " white ", Then combining characters string can be obtained by combination：" whitewood ", " bighead atractylodes rhizome ", it is assumed that " whitewood ", " bighead atractylodes rhizome " hit preset medical word Storehouse, then it may be " whitewood " or " bighead atractylodes rhizome " that can obtain medicine corresponding to the character picture；Assuming that obtained by preceding method Other drugs in the image of prescription to be identified：" radix bupleuri ", " Radix Angelicae Sinensis ", " root of herbaceous peony ", " Poria cocos ", " ginger ", " peppermint ", " process sweet Grass ", then with reference to the property of medicine, " bighead atractylodes rhizome " warm-natured category taste, have the effect of strengthening the spleen and replenishing qi, approached with medicine effect of other several herbal medicines, And " whitewood " this taste medicinal material category liver kidney, cure mainly treating swelling and pain by traumatic injury and other a few taste medicine gaps farther out, thus can consider " bighead atractylodes rhizome " can Energy property is bigger, therefore can select " art ", and abandons " wood ".

In another alternative embodiment of the present invention, include in the attributive character：Treated corresponding to the alternative characters It is described that multiple alternative characters corresponding to each character picture are screened when identifying the medicine similarity of prescription and preset prescription Step A1, can specifically include：Determine that alternative characters correspond to the medicine similarity between prescription to be identified and preset prescription； When the similarity is more than similarity threshold, the preset prescription is chosen from multiple alternative characters corresponding to the character picture Possessed target alternative character.

This alternative embodiment can apply to the situation that most of medicine is fixed in prescription to be identified, describe for convenience, Prescription to be identified can be divided into：Determine medicine and drug candidate, wherein, having determined medicine can be according to identification probability, medicine The attributive character such as vocabulary characteristic obtain, and drug candidate can obtain according to multiple alternative characters corresponding to character picture, then above-mentioned Step A1 can screen to multiple alternative characters corresponding to uncertain medicine.Wherein, if according to multiple corresponding to character picture Multiple combining characters strings that alternative characters and the alternative characters of adjacent character image are formed have hit preset medical dictionary, then this is more Individual combining characters string is corresponding to drug candidate, it is assumed that the quantity of combining characters string is N, then at least two is ordered in N number of combining characters string Preset medical dictionary is suffered.

In actual applications, can be matched medicine has been determined in prescription to be identified with the medicine of preset prescription, in terms of Calculate the medicine similarity of the two.For example, prescription to be identified includes：Determined medicine " radix bupleuri ", " Radix Angelicae Sinensis ", " root of herbaceous peony ", " Poria cocos ", " ginger ", " peppermint ", " honey-fried licorice root " and drug candidate " bighead atractylodes rhizome " and " whitewood ", be computed understanding, the prescription to be identified with it is pre- The medicine similarity for putting prescription " Xiaoyao San " is up to 87%, does not have " whitewood " this taste medicine in " Xiaoyao San " still, and has " white This taste medicine of art ", then can be from drug candidate " bighead atractylodes rhizome " and " whitewood " selection " bighead atractylodes rhizome ".

In another alternative embodiment of the present invention, include in the attributive character：Medicine corresponding to the alternative characters During thing co-occurrence probability, the step A1 that multiple alternative characters corresponding to each character picture are screened, it can specifically wrap Include：Determine character picture to be screened each drug candidate and other character pictures corresponding to medicine co-occurrence probability between medicine； According to the medicine co-occurrence probability, target alternative character is chosen from multiple alternative characters corresponding to the character picture.

Medicine co-occurrence probability can be used for the probability for representing that multi-medicament occurs simultaneously, and its numerical value can be between [0,1]. This provides the implementation process example of the statistical of medicine co-occurrence probability, and above-mentioned statistical can be machine learning mode etc.. For example, whether can often be occurred according to two medical vocabulary in same preset medicine text, to determine two medical words Relation between remittance.Furthermore it is possible to carry out the relation between medical vocabulary according to the relation between a variety of two medical vocabulary Excavate, for example, medical vocabulary A and medical vocabulary B often occur simultaneously, and medical vocabulary B and medical vocabulary C often go out simultaneously It is existing, then it can consider medical vocabulary A and medical vocabulary C also relatively, therefore higher numerical value can be given.It is appreciated that The embodiment of the present invention is not any limitation as the specific acquisition formula of medicine co-occurrence probability.

In actual applications, the medicine co-occurrence probability that a kind of alternative characters are corresponded between drug candidate and other drugs is got over Greatly, it is believed that the possibility that the alternative characters correspond to drug candidate is bigger, on the contrary, it is believed that the alternative characters correspond to alternative The possibility of medicine is smaller, and the selection of target alternative character can be carried out based on this.

It is above-mentioned that multiple alternative characters corresponding to each character picture are sieved in a kind of alternative embodiment of the present invention The step A1 of choosing, can specifically include：According to the medical vocabulary characteristic, medicine property of medicine, the medicine similarity with preset prescription At least one of with medicine co-occurrence probability, identification probability corresponding to corresponding alternative characters is adjusted；After adjustment Identification probability, target alternative character is chosen from multiple alternative characters corresponding to the character picture.For example, in alternative characters , can be to alternative word when the medicine similarity and stronger medicine co-occurrence probability of medical vocabulary characteristic, medicine property of medicine and preset prescription The possibility of symbol is weighted.Otherwise drop power can be carried out to the possibility of alternative characters, identification probability may finally be chosen most Big alternative characters, as target alternative character.Wherein, every kind of attributive character can have corresponding weight, and weighting corresponds to Weight can be more than 1, drop power corresponding to weight can be less than 1, it will be understood that the embodiment of the present invention was for specifically adjusting Journey is not any limitation as.

In actual applications, above-mentioned steps A2 can be according to the sequence of positions in prescription to be identified, to target alternative word Symbol is combined, to obtain corresponding prescription recognition result.Further, client can show prescription identification knot to user Fruit.For example, a kind of example of prescription recognition result can be：Prepared rhizome of rehmannia (1g), the root of herbaceous peony (1g), Radix Angelicae Sinensis (1g), rhizome of chuanxiong (1g).

To sum up, the prescription recognition methods of the embodiment of the present invention, special medical character recognition storehouse can be utilized with medical word The identification for the image that prescription to be identified is carried out for unit is accorded with, because medical character involved in field of medicaments is limited (with the traditional Chinese medical science Exemplified by the identification of prescription, the title of traditional Chinese medical science medicinal material can be enumerated), identify the difficulty of the character picture corresponding to each character Well below the difficulty of the image of the whole prescription to be identified of identification, also, the medical character recognition of medical character can also be directed to Storehouse optimizes, such as a large amount of script samples of single medical character can be used to be trained, therefore can effectively improve Medical character recognition and the accuracy rate of prescription identification.

In addition, in the case of the possible more than one of alternative characters corresponding to above-mentioned character picture, can be according to alternative word Identification probability corresponding to symbol, medical vocabulary characteristic, medicine property of medicine, with the medicine similarity of preset prescription and medicine co-occurrence probability etc. Attributive character determines the possibility of alternative characters, and chooses the maximum alternative characters of possibility as target alternative character, thus The accuracy rate of medical character recognition can be improved.

Embodiment of the method two

Reference picture 2, a kind of step flow chart of prescription recognition methods embodiment two of the present invention is shown, can specifically be wrapped Include following steps：

Step 201, according to character cutting is carried out to the image of prescription to be identified, obtain the character figure corresponding to each character Picture；

Step 202, using medical character recognition storehouse each character picture is identified respectively, obtains each character picture Corresponding alternative characters；The medical character recognition storehouse is got according to the image pattern training of medical character；

Alternative characters corresponding to all character pictures that step 203, the image according to the prescription to be identified are included, are obtained To prescription recognition result corresponding to the image of the prescription to be identified；

Relative to embodiment of the method one shown in Fig. 1, the image of prescription to be identified can specifically include in the present embodiment：It is multiple The image of the prescription to be identified of shooting angle, methods described can also include：

Step 204, foundation prescription recognition result include the quantity of medicine, from the medicine to be identified of the multiple shooting angle Optimal prescription recognition result is chosen in multiple prescription recognition results corresponding to the image of side.

In actual applications, in the image that the image of prescription to be identified voluntarily shoots for user, the figure of prescription to be identified As probably due to angle is not just influenceing the degree of accuracy of recognition result.In view of the above-mentioned problems, the embodiment of the present invention can be directed to it is more The image of the prescription to be identified of individual shooting angle, multiple prescription recognition results are obtained, and medicine is included according to prescription recognition result The quantity of thing, optimal prescription recognition result is chosen from multiple prescription recognition results.As a rule, prescription recognition result is wrapped The more at most corresponding shooting angle of the quantity of drug containing more just, therefore can be chosen and make comprising the most prescription recognition result of medicine For optimal prescription recognition result.

When in the specific implementation, receiving user and being directed to the trigger action of the preset interface, client is taken pictures showing During option, corresponding multi-angle prompt message can be provided by modes such as voice or interfaces, so that user shoots multiple shootings The image of the prescription to be identified of angle and upload.Wherein it is possible to this hair is performed again after the image for uploading multiple prescriptions to be identified The prescription identification process of bright embodiment, or, the present invention can be performed after the image of a prescription to be identified is uploaded and implemented The prescription identification process of example, and prompting user continues to upload the image of next prescription to be identified after the completion of once identifying, so The prescription identification process of the embodiment of the present invention is performed afterwards, it will be understood that the embodiment of the present invention is waited to know for multiple shooting angle The collection of the image of other prescription and recognition sequence are not any limitation as.

To sum up, the embodiment of the present invention includes the quantity of medicine according to prescription recognition result, from the multiple shooting angle Prescription to be identified image corresponding to choose optimal prescription recognition result in multiple prescription recognition results, shooting can be chosen Prescription recognition result corresponding to the most positive image of angle, therefore the degree of accuracy of prescription identification can be improved.

Embodiment of the method three

Reference picture 3, a kind of step flow chart of prescription recognition methods embodiment three of the present invention is shown, can specifically be wrapped Include following steps：

Step 301, according to character cutting is carried out to the image of prescription to be identified, obtain the character figure corresponding to each character Picture；

Step 302, using medical character recognition storehouse each character picture is identified respectively, obtains each character picture Corresponding alternative characters；The medical character recognition storehouse is got according to the image pattern training of medical character；

Alternative characters corresponding to all character pictures that step 303, the image according to the prescription to be identified are included, are obtained To prescription recognition result corresponding to the image of the prescription to be identified；

Relative to embodiment of the method two shown in embodiment of the method one shown in Fig. 1 or Fig. 2, prescription described in the present embodiment is known Other result includes：The title of at least one medicine, methods described can also include：

Mapping relations corresponding to step 304, foundation medicine referred to as between full name, by the prescription recognition result Chinese medicine The abbreviation of the title of thing is converted to full name.

Specifically, judge to whether there is the entitled abbreviation of medicine in the prescription recognition result, if it is present foundation Mapping relations corresponding to medicine referred to as between full name, the abbreviation is converted to the full name of the medicine.

In actual applications, such as in the prescription of doctor trained in Western medicine prescription it there may be that writing is unclear or doctor's conventional characters or list The problem of word abbreviation replaces medicine name.In view of the above-mentioned problems, the embodiment of the present invention can be according to abbreviation and full name corresponding to medicine Between mapping relations, the abbreviation of the title of medicine in the prescription recognition result is converted into full name, and by the full name of medicine User is supplied to, the difficulty of prescription reading can not only be reduced, and the efficiency of prescription typing can be improved.

For example, medicine " Amoxicillin " is referred to as " Amoxil ", some doctors custom uses " Amoxil " in prescription, " Amoxil " can be then utilized to obtain corresponding full name in prescription " Amoxil " with the mapping relations of " Amoxicillin ".

In actual applications, above-mentioned abbreviation can be the simple address or partial character of medicine.Specifically, can be directed to Each medicine, its name is referred to as full name, its corresponding doctor's conventional characters or abbreviation or partial character therein are made For referred to as, with obtain referred to as and full name between mapping relations.So, even if the problem of writing is unclear in prescription be present, also may be used To search to obtain corresponding full name according to the partial character of legibility, or, even if exist in prescription doctor's conventional characters or Word is abridged, and can also search to obtain corresponding full name, therefore can reduce the difficulty of prescription reading.

, can also foundation in a kind of alternative embodiment of the present invention in view of the Problem-Error of drug information in prescription Mapping relations corresponding to medicine between error message and correct information, the title of medicine in the prescription recognition result is carried out Correction process.Wherein, above-mentioned error message, correct information can be corresponding to medicine name, wherein, correct information can be medicine Title, error message can be the wrong usage of common medicine name, for example, the error message of " Amoxil " can be " amoxll ", the error message of " bletilla " can be " bletilla striata ", and the error message of " coptis " can be " coptis root ", " Herba Andrographitis " Error message can be " punching company " etc..

It should be noted that can by the mapping relations corresponding to said medicine between error message and correct information, with Referred to as it is used in combination corresponding to medicine with the mapping relations between full name.For example, the error correction of abbreviation can be carried out first, then will Correctly referred to as be converted to full name, first by " amoxll " be corrected as " amoxil " then by " amoxil " be converted to Chinese " Ah Amdinocillin " etc..

In a kind of alternative embodiment of the present invention, in addition to the full name of medicine is provided, medicine can also be provided a user Relevant information.By taking Western medicine medicine as an example, its corresponding relevant information can specifically include：Trade name, common name, specification, life Produce the information such as producer, batch number, authentication code, major function, taboo, adverse reaction and shelf-life.By taking Chinese medicine medicine as an example, Its corresponding relevant information can specifically include：Title, the place of production, specification, shelf-life, authentication code etc..By taking traditional Chinese medical science medicinal material as an example, Its corresponding relevant information can specifically include：The information such as morphological feature, geographical distribution, value, discriminating.

Embodiment of the method four

Reference picture 4, a kind of step flow chart of prescription recognition methods example IV of the present invention is shown, can specifically be wrapped Include following steps：

Step 401, according to separator cutting is carried out to the image of prescription to be identified, obtain the medicine corresponding to every kind of medicine Character picture；

In actual applications, it can include being used for splitting the separator of medicine in prescription to be identified, such as ", ", ", ", "：”、 " (", ") ", space etc., it can identify to obtain above-mentioned separator using image recognition technology, and according to separator to medicine to be identified The image of side carries out cutting, to obtain the medicine character picture in units of medicine.

Step 402, in units of character, cutting is carried out to the medicine character picture, obtained corresponding to each character single Character picture；

Step 403, using medical character recognition storehouse each monocase image is identified respectively, obtains each monocase Alternative characters corresponding to image；The medical character recognition storehouse is got according to the image pattern training of medical character；

Step 404, for each medicine character picture, the alternative characters of all monocase images that it is included are carried out Combination, with combining characters string corresponding to composition；

Specifically, the alternative characters between adjacent monocase image that each medicine character picture is included can be carried out Combination, with combining characters string corresponding to composition.Wherein, when certain monocase image corresponds to multiple alternative characters, can obtain more Individual combining characters string.For example, a medicine character picture includes 2 monocase images, wherein, first monocase image is corresponding Alternative characters be " white ", alternative characters corresponding to second monocase image include：" not " and " art ", then can respectively by " not " and " art " forms combining characters string with " white " respectively：" in vain not " and " bighead atractylodes rhizome ".

Step 405, when medicine character picture corresponds to multiple combining characters strings, for each combining characters string, judge pre- Put in medical dictionary with the presence or absence of the medical vocabulary for including the combining characters string；If it is present judge to hit preset medical word Whether the number of the combining characters string in storehouse is more than 1, if, it is determined that it is drug candidate that medicine character picture, which corresponds to medicine, and is held Row step 406 either step 407 or step 408, otherwise, it determines medicine character picture corresponds to medicine to have determined medicine；

When multiple combining characters strings hit preset medical dictionary corresponding to the medicine character picture, step can be passed through Rapid 406, at least one of step 407 and step 408, the multiple combining characters strings for hitting preset medical dictionary are sieved Choosing, to obtain corresponding objective cross character string.

Step 406, when multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, according to each Combining characters string corresponds to the property of medicine diversity factor between medicine corresponding to drug candidate and other drugs character picture, from the multiple The preset prescription possessed objective cross character string is chosen in combining characters string；

Assuming that combining characters string corresponding to certain medicine character picture is：" whitewood " and " bighead atractylodes rhizome ", and obtained by preceding method The other drugs into the image of prescription to be identified：" radix bupleuri ", " Radix Angelicae Sinensis ", " root of herbaceous peony ", " Poria cocos ", " ginger ", " peppermint ", " process Radix glycyrrhizae ", then with reference to the property of medicine, " bighead atractylodes rhizome " warm-natured category taste, there is the effect of strengthening the spleen and replenishing qi, connect with medicine effect of other several herbal medicines Closely, " whitewood " this taste medicinal material category liver kidney, treating swelling and pain by traumatic injury and other a few taste medicine gaps are cured mainly farther out, therefore can consider " bighead atractylodes rhizome " Possibility is bigger, therefore can select " bighead atractylodes rhizome ", and abandons " whitewood ".

It should be noted that to hit preset medical dictionary specific for multiple combining characters strings corresponding to medicine character picture Refer to, have in combining characters string corresponding to medicine character picture it is multiple hit preset medical dictionary, for example, medicine character picture pair P combining characters string is answered, then hits preset medical dictionary in whole or in part in the P combining characters string, P is more than 1 Integer.

Step 407, when multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, it is determined that respectively Combining characters string corresponds to drug candidate and has determined the medicine similarity corresponding to medicine between prescription to be identified and preset prescription； When the similarity is more than similarity threshold, the preset prescription possessed target is chosen from the multiple combining characters string Combining characters string；

For example, prescription to be identified includes：Medicine " radix bupleuri ", " Radix Angelicae Sinensis ", " root of herbaceous peony ", " Poria cocos ", " ginger ", " thin are determined Lotus ", " honey-fried licorice root " and drug candidate " bighead atractylodes rhizome " and " whitewood ", it is computed understanding, the prescription to be identified and preset prescription are " free and unfettered It is distant to dissipate " medicine similarity be up to 87%, do not have " whitewood " this taste medicine in " Xiaoyao San " still, and have " bighead atractylodes rhizome " this taste medicine, Then can be from drug candidate " bighead atractylodes rhizome " and " whitewood " selection " bighead atractylodes rhizome ".

Step 408, when multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, it is determined that respectively Combining characters string corresponds to the medicine co-occurrence probability between medicine corresponding to drug candidate and other character pictures；According to the medicine Co-occurrence probability, objective cross character string is chosen from excessively individual combining characters string；

It should be noted that step 406, step 407 or step 408 choose target from the multiple combining characters string The identification probability of alternative characters can be used during combining characters string, also, step 406, step 407 and step 408 can To be used in combination.

Step 409, at least one of step 406, step 407 and step 408 output objective cross character string and its Combining characters string corresponding to his medicine character picture is combined, to obtain corresponding prescription recognition result.

In actual applications, sequence of positions that can be according to medicine character picture in prescription to be identified, to combining characters String is combined, to obtain corresponding prescription recognition result.For example, a kind of example of prescription recognition result can be：Prepared rhizome of rehmannia (1g), the root of herbaceous peony (1g), Radix Angelicae Sinensis (1g), rhizome of chuanxiong (1g).

To sum up, the prescription recognition methods of the embodiment of the present invention, special medical character recognition storehouse can be utilized with medical word The identification for the monocase image that prescription to be identified is carried out for unit is accorded with, because medical character involved in field of medicaments is limited (by taking the identification of traditional Chinese medical science prescription as an example, the title of traditional Chinese medical science medicinal material can be enumerated), identifies the character figure corresponding to each character The difficulty of picture can also be directed to the medicine of medical character well below the difficulty for identifying the entirely image of prescription to be identified Character recognition storehouse optimizes, such as a large amount of script samples of single medical character can be used to be trained, therefore can Effectively improve the accuracy rate of medical character recognition and prescription identification.

In addition, after alternative characters corresponding to each monocase image are obtained, can also be wrapped according to medicine character picture Alternative characters containing all monocase images form combining characters string, and in units of medicine, combining characters string is screened, Therefore the reasonability of combining characters string screening can be improved, so as to effectively improve the standard of medicine character recognition and prescription identification True rate.

It should be noted that for embodiment of the method, in order to be briefly described, therefore it is dynamic that it is all expressed as to a series of motion Combine, but those skilled in the art should know, the embodiment of the present invention is not limited by described athletic performance order System, because according to the embodiment of the present invention, some steps can use other orders or carry out simultaneously.Secondly, art technology Personnel should also know that embodiment described in this description belongs to preferred embodiment, and involved athletic performance simultaneously differs Surely necessary to being the embodiment of the present invention.

Device embodiment

Reference picture 5, a kind of structured flowchart of prescription identification device embodiment of the present invention is shown, can specifically be included： Image cutting module 501, picture recognition module 502 and result acquisition module 503.

Image cutting module 501, for carrying out cutting to the image of prescription to be identified according to character, obtain each character institute Corresponding character picture；

Picture recognition module 502, for each character picture to be identified respectively using medical character recognition storehouse, obtain Alternative characters corresponding to each character picture；The medical character recognition storehouse can be according to the image pattern collection training of medical character Obtain；And

As a result acquisition module 503, all character pictures for being included according to the image of the prescription to be identified are corresponding Alternative characters, obtain prescription recognition result corresponding to the image of the prescription to be identified.

Alternatively, when each character picture corresponds to multiple alternative characters, the result acquisition module 503 can specifically wrap Include：

Alternatively, the attributive character specifically includes：Medical vocabulary characteristic corresponding to the alternative characters, then it is described alternative Character screening submodule can specifically include：

Alternatively, the determining unit can specifically include：

Alternatively, the attributive character can include：Medicine property of medicine corresponding to the alternative characters, then alternative characters Screening submodule can specifically include：

Alternatively, the attributive character can include：The corresponding medicine similarity with preset prescription of the alternative characters, Then the alternative characters screening submodule can specifically include：

Alternatively, the attributive character can include：Medicine co-occurrence probability corresponding to the alternative characters, then it is described alternative Character screening submodule can include：

Alternatively, the alternative characters screening submodule can include：

Alternatively, the image of the prescription to be identified can specifically include：The figure of the prescription to be identified of multiple shooting angle Picture, then described device can also include：

Alternatively, the prescription recognition result can specifically include：The title of at least one medicine, then described device may be used also With including：

Alternatively, above-mentioned image cutting module 501 can specifically include：

Alternatively, the above results acquisition module 503 can specifically include：

Alternatively, combinations thereof character string selection submodule can specifically include：

For device embodiment, because it is substantially similar to embodiment of the method, so description is fairly simple, it is related Part illustrates referring to the part of embodiment of the method.

Each embodiment in this specification is described by the way of progressive, what each embodiment stressed be with The difference of other embodiment, between each embodiment identical similar part mutually referring to.

On the device in above-described embodiment, wherein modules perform the concrete mode of operation in relevant this method Embodiment in be described in detail, explanation will be not set forth in detail herein.

Fig. 6 is a kind of block diagram of device 900 for prescription identification according to an exemplary embodiment.For example, dress It can be mobile phone to put 900, computer, digital broadcast terminal, messaging devices, game console, tablet device, medical treatment Equipment, body-building equipment, personal digital assistant etc..

Reference picture 6, device 900 can include following one or more assemblies：Processing component 902, memory 904, power supply Component 906, multimedia groupware 908, audio-frequency assembly 910, the interface 912 of input/output (I/O), sensor cluster 914, and Communication component 916.

The integrated operation of the usual control device 900 of processing component 902, such as communicated with display, call, data, phase The operation that machine operates and record operation is associated.Treatment element 902 can refer to including one or more processors 930 to perform Order, to complete all or part of step of above-mentioned method.In addition, processing component 902 can include one or more modules, just Interaction between processing component 902 and other assemblies.For example, processing component 902 can include multi-media module, it is more to facilitate Interaction between media component 908 and processing component 902.

Memory 904 is configured as storing various types of data to support the operation in equipment 900.These data are shown Example includes the instruction of any application program or method for being operated on device 900, contact data, telephone book data, disappears Breath, picture, video etc..Memory 904 can be by any kind of volatibility or non-volatile memory device or their group Close and realize, as static RAM (SRAM), Electrically Erasable Read Only Memory (EEPROM) are erasable to compile Journey read-only storage (EPROM), programmable read only memory (PROM), read-only storage (ROM), magnetic memory, flash Device, disk or CD.

Power supply module 906 provides electric power for the various assemblies of device 900.Power supply module 906 can include power management system System, one or more power supplys, and other components associated with generating, managing and distributing electric power for device 900.

Multimedia groupware 908 is included in the screen of one output interface of offer between described device 900 and user.One In a little embodiments, screen can include liquid crystal display (LCD) and touch panel (TP).If screen includes touch panel, screen Curtain may be implemented as touch-screen, to receive the input signal from user.Touch panel includes one or more touch sensings Device is with the gesture on sensing touch, slip and touch panel.The touch sensor can not only sensing touch or sliding motion The border of action, but also detect the duration and pressure related to the touch or slide.In certain embodiments, Multimedia groupware 908 includes a front camera and/or rear camera.When equipment 900 is in operator scheme, mould is such as shot When formula or video mode, front camera and/or rear camera can receive outside multi-medium data.Each preposition shooting Head and rear camera can be a fixed optical lens system or have focusing and optical zoom capabilities.

Audio-frequency assembly 910 is configured as output and/or input audio signal.For example, audio-frequency assembly 910 includes a Mike Wind (MIC), when device 900 is in operator scheme, during such as call model, logging mode and speech recognition mode, microphone by with It is set to reception external audio signal.The audio signal received can be further stored in memory 904 or via communication set Part 916 is sent.In certain embodiments, audio-frequency assembly 910 also includes a loudspeaker, for exports audio signal.

I/O interfaces 912 provide interface between processing component 902 and peripheral interface module, and above-mentioned peripheral interface module can To be keyboard, click wheel, button etc..These buttons may include but be not limited to：Home button, volume button, start button and lock Determine button.

Sensor cluster 914 includes one or more sensors, and the state for providing various aspects for device 900 is commented Estimate.For example, sensor cluster 914 can detect opening/closed mode of equipment 900, and the relative positioning of component, for example, it is described Component is the display and keypad of device 900, and sensor cluster 914 can be with 900 1 components of detection means 900 or device Position change, the existence or non-existence that user contacts with device 900, the orientation of device 900 or acceleration/deceleration and device 900 Temperature change.Sensor cluster 914 can include proximity transducer, be configured to detect in no any physical contact The presence of neighbouring object.Sensor cluster 914 can also include optical sensor, such as CMOS or ccd image sensor, for into As being used in application.In certain embodiments, the sensor cluster 914 can also include acceleration transducer, gyro sensors Device, Magnetic Sensor, pressure sensor or temperature sensor.

Communication component 916 is configured to facilitate the communication of wired or wireless way between device 900 and other equipment.Device 900 can access the wireless network based on communication standard, such as WiFi, 2G or 3G, or combinations thereof.In an exemplary implementation In example, communication component 916 receives broadcast singal or broadcast related information from external broadcasting management system via broadcast channel. In one exemplary embodiment, the communication component 916 also includes near-field communication (NFC) module, to promote junction service.Example Such as, in NFC module radio frequency identification (RFID) technology can be based on, Infrared Data Association (IrDA) technology, ultra wide band (UWB) technology, Bluetooth (BT) technology and other technologies are realized.

In the exemplary embodiment, device 900 can be believed by one or more application specific integrated circuits (ASIC), numeral Number processor (DSP), digital signal processing appts (DSPD), PLD (PLD), field programmable gate array (FPGA), controller, microcontroller, microprocessor or other electronic components are realized, for performing the above method.

In the exemplary embodiment, a kind of non-transitorycomputer readable storage medium including instructing, example are additionally provided Such as include the memory 604 of instruction, above-mentioned instruction can be performed to complete the above method by the processor 630 of device 600.For example, The non-transitorycomputer readable storage medium can be ROM, random access memory (RAM), CD-ROM, tape, floppy disk With optical data storage devices etc..

A kind of non-transitorycomputer readable storage medium, when the instruction in the storage medium is by the processing of intelligent terminal When device performs so that intelligent terminal is able to carry out a kind of prescription recognition methods, and methods described includes：According to character to medicine to be identified The image of side carries out cutting, obtains the character picture corresponding to each character；Medical character recognition storehouse is respectively to each character figure As being identified, alternative characters corresponding to each character picture are obtained；Figure of the medical character recognition storehouse according to medical character Decent training is got；Alternative characters corresponding to all character pictures that image according to the prescription to be identified is included, Obtain prescription recognition result corresponding to the image of the prescription to be identified.

Fig. 7 is the structural representation of server in some embodiments of the present invention.The server 1900 can be because of configuration or property Energy is different and produces bigger difference, can include one or more central processing units (central processing Units, CPU) 1922 (for example, one or more processors) and memory 1932, one or more storage applications The storage medium 1930 of program 1942 or data 1944 (such as one or more mass memory units).Wherein, memory 1932 and storage medium 1930 can be it is of short duration storage or persistently storage.One can be included by being stored in the program of storage medium 1930 Individual or more than one module (diagram does not mark), each module can include operating the series of instructions in server.More enter One step, central processing unit 1922 be could be arranged to communicate with storage medium 1930, and storage medium is performed on server 1900 Series of instructions operation in 1930.

Server 1900 can also include one or more power supplys 1926, one or more wired or wireless nets Network interface 1950, one or more input/output interfaces 1958, one or more keyboards 1956, and/or, one or More than one operating system 1941, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM Etc..

Those skilled in the art will readily occur to the present invention its after considering specification and putting into practice invention disclosed herein Its embodiment.It is contemplated that cover the present invention any modification, purposes or adaptations, these modifications, purposes or Person's adaptations follow the general principle of the present invention and including the undocumented common knowledges in the art of the disclosure Or conventional techniques.Description and embodiments are considered only as exemplary, and true scope and spirit of the invention are by following Claim is pointed out.

It should be appreciated that the invention is not limited in the precision architecture for being described above and being shown in the drawings, and And various modifications and changes can be being carried out without departing from the scope.The scope of the present invention is only limited by appended claim

The foregoing is only presently preferred embodiments of the present invention, be not intended to limit the invention, it is all the present invention spirit and Within principle, any modification, equivalent substitution and improvements made etc., it should be included in the scope of the protection.

A kind of prescription recognition methods provided by the present invention, a kind of prescription identification device and a kind of prescription that is used for are known above Other device, is described in detail, and specific case used herein is explained the principle and embodiment of the present invention State, the explanation of above example is only intended to help the method and its core concept for understanding the present invention；Meanwhile for this area Those skilled in the art, according to the thought of the present invention, there will be changes in specific embodiments and applications, to sum up institute State, this specification content should not be construed as limiting the invention.

Claims

A kind of 1. prescription recognition methods, it is characterised in that including：

Cutting is carried out to the image of prescription to be identified according to character, obtains the character picture corresponding to each character；

Each character picture is identified respectively using medical character recognition storehouse, obtains alternative word corresponding to each character picture Symbol；The medical character recognition storehouse is got according to the image pattern training of medical character；

Alternative characters corresponding to all character pictures that image according to the prescription to be identified is included, obtain described to be identified Prescription recognition result corresponding to the image of prescription.
2. according to the method for claim 1, it is characterised in that when each character picture corresponds to multiple alternative characters, institute Alternative characters corresponding to stating all character pictures included according to the image of the prescription to be identified, obtain the medicine to be identified Corresponding to the image of side the step of prescription recognition result, including：

The attributive character of alternative characters, right corresponding to each character picture included in image according to the prescription to be identified Multiple alternative characters corresponding to each character picture are screened, to obtain target alternative character corresponding to each character picture；

Target alternative character corresponding to all character pictures difference included in image according to the prescription to be identified, obtains institute State prescription recognition result corresponding to the image of prescription to be identified；

Wherein, the attributive character includes at least one of following feature：Identification probability, medicine corresponding to the alternative characters Vocabulary characteristic, medicine property of medicine, medicine similarity and medicine co-occurrence probability with preset prescription.
3. according to the method for claim 2, it is characterised in that the attributive character includes：Corresponding to the alternative characters Medical vocabulary characteristic, then it is described to the step of multiple alternative characters are screened corresponding to each character picture, including：

For multiple alternative characters corresponding to each character picture, determine that each character picture is corresponding using preset medical dictionary Target alternative character.
4. according to the method for claim 3, it is characterised in that described that each character figure is determined using preset medical dictionary As corresponding to the step of target alternative character, including：

It is for each character picture, multiple alternative characters corresponding to the character picture are corresponding with adjacent character image multiple Alternative characters, form multiple combining characters strings；

For each combining characters string, judge in the preset medical dictionary with the presence or absence of the medicine for including the combining characters string Vocabulary；If it is present using alternative characters corresponding to the combining characters string for hitting preset medical dictionary as the character picture Corresponding target alternative character.
5. according to the method for claim 3, it is characterised in that described that each character figure is determined using preset medical dictionary As corresponding to the step of target alternative character, including：

It is for each character picture, multiple alternative characters corresponding to the character picture are corresponding with adjacent character image multiple Alternative characters, form multiple combining characters strings；

For each combining characters string, judge in the preset medical dictionary with the presence or absence of the medicine for including the combining characters string Vocabulary；If it is present the identification probability of alternative characters corresponding to the combining characters string is weighted；

All combining characters strings are judged after terminating, select the maximum alternative characters of identification probability corresponding as the character picture Target alternative character.
6. according to the method for claim 2, it is characterised in that the attributive character includes：Corresponding to the alternative characters Medicine property of medicine, then it is described to the step of multiple alternative characters are screened corresponding to each character picture, including：

Image according to prescription to be identified includes the property of medicine that character picture corresponds to medicine, determines each standby of character picture to be screened The property of medicine diversity factor between medicine corresponding to medicine and other character pictures is selected, wherein, medicine corresponding to the character picture is Combination according to alternative characters corresponding to adjacent character image obtains；

According to the property of medicine diversity factor, target alternative character is chosen from multiple alternative characters corresponding to character picture to be screened.
7. according to the method for claim 2, it is characterised in that the attributive character includes：Corresponding to the alternative characters It is with the medicine similarity of preset prescription, then described to the step of multiple alternative characters are screened corresponding to each character picture, Including：

Determine that alternative characters correspond to the medicine similarity between prescription to be identified and preset prescription；

When the similarity is more than similarity threshold, chosen from multiple alternative characters corresponding to the character picture described pre- Put prescription possessed target alternative character.
8. according to the method for claim 2, it is characterised in that the attributive character includes：Corresponding to the alternative characters Medicine co-occurrence probability, it is described to the step of multiple alternative characters are screened corresponding to each character picture, including：

Determine character picture to be screened each drug candidate and other character pictures corresponding to medicine co-occurrence probability between medicine；

According to the medicine co-occurrence probability, target alternative character is chosen from multiple alternative characters corresponding to the character picture.
9. according to the method for claim 2, it is characterised in that described to multiple alternative characters corresponding to each character picture The step of being screened, including：

According to the medical vocabulary characteristic, medicine property of medicine, with the medicine similarity of preset prescription and medicine co-occurrence probability extremely Few one kind, is adjusted to identification probability corresponding to corresponding alternative characters；

According to the identification probability after adjustment, target alternative character is chosen from multiple alternative characters corresponding to the character picture.
10. according to any described method in claim 1 to 9, it is characterised in that the image of the prescription to be identified includes： The image of the prescription to be identified of multiple shooting angle, methods described also include：

The quantity of medicine is included according to prescription recognition result, it is corresponding from the image of the prescription to be identified of the multiple shooting angle Multiple prescription recognition results in choose optimal prescription recognition result.
11. according to any described method in claim 1 to 9, it is characterised in that the prescription recognition result includes：At least A kind of title of medicine, methods described also include：

According to the mapping relations corresponding to medicine referred to as between full name, by the letter of the title of medicine in the prescription recognition result Title is converted to full name.
12. according to the method for claim 1, it is characterised in that described that the image of prescription to be identified is carried out according to character Cutting, the character picture corresponding to each character is obtained, including：

Cutting is carried out to the image of prescription to be identified according to separator, obtains the medicine character picture corresponding to every kind of medicine；

In units of character, cutting is carried out to the medicine character picture, obtains monocase image corresponding to each character.
13. according to the method for claim 12, it is characterised in that the image according to the prescription to be identified is included All character pictures corresponding to alternative characters, obtain prescription recognition result corresponding to the image of the prescription to be identified, including：

For each medicine character picture, the alternative characters for all monocase images that it is included are combined, to form Corresponding combining characters string；

When the medicine character picture corresponds to multiple combining characters strings, for each combining characters string, preset medical word is judged With the presence or absence of the medical vocabulary for including the combining characters string in storehouse；

When multiple combining characters strings hit preset medical dictionary corresponding to the medicine character picture, to hitting preset medicine Multiple combining characters strings of dictionary are screened, to obtain corresponding objective cross character string；

By combining characters string corresponding to objective cross character string corresponding to the medicine character picture and other drugs character picture It is combined, to obtain corresponding prescription recognition result.
14. according to the method for claim 13, it is characterised in that multiple combined characters of the described pair of preset medical dictionary of hit The step of symbol string is screened, including：

When multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, according to each combination character string pair The property of medicine diversity factor between medicine corresponding to drug candidate and other drugs character picture is answered, from the multiple combining characters string Choose objective cross character string；And/or

When multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, determine that each combination character string is standby Select medicine and determine the medicine similarity corresponding to medicine between prescription to be identified and preset prescription；It is more than phase in the similarity During like degree threshold value, the preset prescription possessed objective cross character string is chosen from the multiple combining characters string；With/ Or,

When multiple combining characters strings hit preset medical dictionary corresponding to medicine character picture, each combination character string pair is determined Answer the medicine co-occurrence probability between medicine corresponding to drug candidate and other character pictures；According to the medicine co-occurrence probability, from Objective cross character string is chosen in the multiple combining characters string.
A kind of 15. prescription identification device, it is characterised in that including：

Image cutting module, for carrying out cutting to the image of prescription to be identified according to character, obtain corresponding to each character Character picture；

Picture recognition module, for each character picture to be identified respectively using medical character recognition storehouse, obtain each word Accord with alternative characters corresponding to image；The medical character recognition storehouse is got according to the image pattern training of medical character；And

As a result acquisition module, for alternative word corresponding to all character pictures for being included according to the image of the prescription to be identified Symbol, obtains prescription recognition result corresponding to the image of the prescription to be identified.
A kind of 16. device for prescription identification, it is characterised in that include memory, and one or more than one Program, one of them or more than one program storage are configured to by one or more than one processing in memory Device performs one or more than one program bag and contains the instruction for being used for being operated below：

Cutting is carried out to the image of prescription to be identified according to character, obtains the character picture corresponding to each character；

Each character picture is identified respectively using medical character recognition storehouse, obtains alternative word corresponding to each character picture Symbol；The medical character recognition storehouse is got according to the image pattern training of medical character；

Alternative characters corresponding to all character pictures that image according to the prescription to be identified is included, obtain described to be identified Prescription recognition result corresponding to the image of prescription.