[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

WO2012016484A1 - 有价文件识别方法及其识别系统、装置 - Google Patents

有价文件识别方法及其识别系统、装置 Download PDF

Info

Publication number
WO2012016484A1
WO2012016484A1 PCT/CN2011/076550 CN2011076550W WO2012016484A1 WO 2012016484 A1 WO2012016484 A1 WO 2012016484A1 CN 2011076550 W CN2011076550 W CN 2011076550W WO 2012016484 A1 WO2012016484 A1 WO 2012016484A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
banknote
value document
feature
module
Prior art date
Application number
PCT/CN2011/076550
Other languages
English (en)
French (fr)
Inventor
梁添才
牟总斌
肖大海
Original Assignee
广州广电运通金融电子股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 广州广电运通金融电子股份有限公司 filed Critical 广州广电运通金融电子股份有限公司
Priority to AU2011288069A priority Critical patent/AU2011288069B2/en
Priority to US13/810,422 priority patent/US9262677B2/en
Priority to EP11814067.2A priority patent/EP2602771A4/en
Publication of WO2012016484A1 publication Critical patent/WO2012016484A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10Character recognition
    • G06V30/19Recognition using electronic means
    • G06V30/191Design or setup of recognition systems or techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
    • G06V30/19173Classification techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • G06F18/241Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06F18/24133Distances to prototypes
    • G06F18/24137Distances to cluster centroïds
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/10Image acquisition
    • G06V10/12Details of acquisition arrangements; Constructional details thereof
    • G06V10/14Optical characteristics of the device performing the acquisition or on the illumination arrangements
    • G06V10/143Sensing or illuminating at different wavelengths
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/25Determination of region of interest [ROI] or a volume of interest [VOI]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/30Noise filtering
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/44Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/40Document-oriented image-based pattern recognition
    • G06V30/41Analysis of document content
    • G06V30/418Document matching, e.g. of document images
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/20Testing patterns thereon
    • G07D7/2008Testing patterns thereon using pre-processing, e.g. de-blurring, averaging, normalisation or rotation
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/20Testing patterns thereon
    • G07D7/202Testing patterns thereon using pattern matching
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/20Testing patterns thereon
    • G07D7/202Testing patterns thereon using pattern matching
    • G07D7/206Matching template patterns

Definitions

  • the present invention relates to image processing and pattern recognition techniques, and more particularly to value documents, such as banknote identification methods, and identification systems and apparatus therefor.
  • Image restoration technology based on partial differential equations has achieved certain research results. It can deal with the phenomenon of scratches, handwriting, stains, wrinkles, etc., and can achieve better recovery effect. The recovery of old banknote images and partial differential equations. The research content of image restoration is consistent. Image restoration technology based on partial differential equations is applied to banknote image recovery, which can achieve better recovery results.
  • Fig. 1 is a flow chart of the conventional banknote image recognition processing. As shown in the figure, the image is restored to the entire image, and the features are extracted, and then classified according to the currency, denomination or new and old standards, and finally the authentication is recognized.
  • T denote the time of each banknote identification
  • Represents the time overhead of image processing
  • represents the time overhead of feature extraction
  • represents the time overhead of the classification
  • represents the time overhead of the authentication
  • the existing banknote recognition system does not process the old banknotes, that is, refuses to identify the old banknotes, thereby causing a large number of old banknotes to be rejected, and the banknote identification system has a high rate of banknote rejection, which affects the banknote identification system. Ease of use is not conducive to the promotion and application of financial self-service devices.
  • An object of the present invention is to provide a value document identification method with a fast real-time response speed, and a value document recognition system and apparatus having a fast real-time response speed and a low banknote rejection rate.
  • the method of identifying valuable documents includes:
  • Step 1 extracting features, selecting a feature area of the value document, and extracting a value document feature for quick classification from the feature area;
  • Step 2 Quickly classify, according to the value document feature extracted in step 1, the classified value file is quickly sorted according to a preset classification model, and the currency, denomination, orientation, and image quality information of the value document are obtained. Information, and select banknotes with good image quality, one banknote and one banknote with poor image quality - old banknotes;
  • Step 3 image restoration, applying image restoration technology based on partial differential equation to recover the old banknote image, and obtaining the restored image;
  • Step 4 Anti-counterfeiting processing, directly identifying the new banknotes; authenticating the old banknotes through the restored images to determine the authenticity of the current banknotes;
  • Step 5 Output the result of determining the current banknote.
  • the extracting feature in step 1 is to combine the multi-wavelength image characteristics of the banknote, the multi-resolution characteristic of the banknote image, and the anti-counterfeiting property of the banknote image, select the feature region of the multi-wavelength image of the banknote, and combine the classification model to select from Features are extracted from the feature area.
  • the value document feature in the step 1 includes at least one of a white watermark, a black watermark, a color changing ink, a magnetic security thread, a microtext, a gravure printing pattern, a denomination digital number, and a banknote crown number.
  • the vectorization of the feature of the value document in the step 1 includes: normalizing the image of the feature area of the value document to obtain a normalized image; and selecting "the normalized image" Different feature regions are respectively subjected to gray value average calculation for 1 to n feature region images; feature vectors are formed according to the above 1 to n gray scale average values.
  • the quick classification model is established according to different currencies, denominations, orientations, and image quality.
  • the fast classification model includes the following levels:
  • the first layer different currencies, such as: RMB, Euro, US dollar;
  • Second level different denominations in the same currency
  • Third layer different orientations of the same denomination
  • Fourth layer Different image quality of the same denomination, including new and old notes.
  • classification model may be in the following hierarchical order:
  • the first layer different currencies, such as: RMB, Euro, US dollar;
  • Second level different denominations in the same currency
  • the third layer different image quality of the same denomination, including new and old notes;
  • the fourth layer different orientations of the same denomination.
  • the classification model is established by the following steps: dividing p currencies, q class denominations of each currency, front and back 2 faces of each denomination, and feature regions of new and old images; respectively, forming the previous step ⁇ ⁇ ⁇ 4 different images are normalized to obtain a normalized image; from each normalized image, "a different feature region is selected, and for each of the 1 to n feature region images, the gray value average calculation is performed, Forming feature vectors; training the feature vectors by R training samples to form 4 cluster centers of pxgx, and then establishing a classification model of the feature areas of the value documents.
  • the rapid classification process includes the following steps: For extracted features, the distance from the cluster center of each category is:
  • the input D when [0.1-0.5], the input D can be classified into classes; otherwise, the input D does not belong to the category.
  • step 3 the image is restored, and the restored object is a region of interest (R0I) of the banknotes with poor image quality.
  • the restoration of the old banknote image is performed using an image restoration method based on a partial differential equation.
  • the value document identification system provided by the invention comprises:
  • An image acquisition module configured to collect image data of a value document
  • a storage module configured to store the foregoing image data and required standard model data
  • a data processing module performs feature region feature extraction, fast classification, and image restoration processing on the stored image data; that is, selecting a feature region of the value document, and extracting features of the value document from the feature region; Describe the characteristics of the value document, perform rapid classification processing on the input value document according to the pre-set classification model, select banknotes with good image quality and banknotes with poor image quality; and recover the image of the banknote with poor image quality, Obtaining the restored image (ie, image data);
  • An identification module configured to compare the processed image data with the standard model data, and perform an authentication process
  • An output module configured to output an identification result of the identification module
  • a central control module for controlling and coordinating work between the above modules (ie, image acquisition module, storage module, data processing module, identification module, and output module), including data transmission, resource allocation, and the like.
  • a data link is formed between each module in the value document identification system, wherein the data link belongs to the public Known technology, the present invention will not be described again.
  • the invention provides a value document identification device, comprising:
  • a selection module for selecting a feature area of the value document
  • An extraction module configured to extract a feature of the value document from the feature area
  • a classification processing module configured to perform fast classification processing on the input value document according to the extracted classification document according to a preset classification model, and select a banknote with good image quality and a banknote with poor image quality;
  • the identification module is configured to perform the forensic identification of the banknotes with good image quality obtained by the classification processing module; and perform the forgery identification of the image recovered by the recovery processing module to determine the authenticity of the current banknote;
  • An output module configured to output a result of the identification module determining the current banknote.
  • the beneficial effects of the value document identification method, device and system are: omitting the image restoration processing of the banknotes with good image quality, saving time, improving the processing efficiency of the whole system; and having poor image quality Banknotes - the area of interest (ROI) of the old banknotes is restored, which not only saves system overhead, but also provides good quality image data for the authentication process, reduces the difficulty of authenticity, and improves the accuracy of the authentication, thereby improving the banknote identification system. Receive rate.
  • ROI area of interest
  • FIG. 1 is a flow chart of a conventional banknote image recognition processing method
  • FIG. 2 is a flow chart of a method for identifying a value document provided by an embodiment of the present invention
  • FIG. 3 is a schematic diagram of establishing a rapid classification model in FIG. 2;
  • Figure 4 is another schematic diagram of establishing a quick classification in Figure 2;
  • Figure 5 is a schematic diagram of image recovery in Figure 2;
  • FIG. 6 is a schematic structural diagram of a value document identification system according to an embodiment of the present invention.
  • the value document identification method includes: Step 1: extracting features; Step 2, fast classification; Step 3, image restoration; Step 4, authentication identification; and Step 5, outputting the result.
  • Step 1 extracting features
  • Step 2, fast classification fast classification
  • Step 3, image restoration image restoration
  • Step 4, authentication identification authentication identification
  • Step 5 outputting the result.
  • the details of each step are detailed below.
  • Step 1 Extract features.
  • the feature area of the multi-wavelength image of the banknote is selected, and the feature is extracted from the selected feature area in combination with the classification model, such as: , black watermarks, color-changing inks, magnetic security lines, micro-text, gravure prints, denomination digit numbers, and/or banknote crown numbers.
  • the classification model such as: , black watermarks, color-changing inks, magnetic security lines, micro-text, gravure prints, denomination digit numbers, and/or banknote crown numbers.
  • the feature area image is normalized to obtain a normalized image:
  • Step 2 Quickly categorize.
  • the input banknote information is quickly sorted according to a predetermined classification model to obtain the currency, denomination, orientation, and image quality of the banknote.
  • the currency, denomination, and orientation image quality provide guidelines for subsequent ring-orientation interest (ROI) regions.
  • ROI ring-orientation interest
  • the image quality determines whether image restoration processing is required, and the banknotes with good image quality are directly sent to the recognition. Processing, old banknotes with poor image quality are processed by the data processing module.
  • First, the classification model is established according to different currencies, denominations and image quality in the following levels: First layer: According to different currencies, such as: RMB, Euro, US dollar, the banknotes are classified; Second layer: According to the same currency Different denominations, such as: RMB 100, 50 yuan, further classification of banknotes;
  • the third layer according to the different orientation of the same denomination, such as: forward, reverse, the third level of the banknote classification; then
  • the fourth layer According to different image quality of the same denomination, such as: new banknotes, old banknotes, classify the banknotes again.
  • the third layer and the fourth layer can be interchanged, as shown in Figure 4.
  • the first layer according to different currencies, such as: RMB, Euro, US dollar, classify banknotes;
  • Second layer According to the same currency Different denominations, such as: RMB 100, 50 yuan, the second level of banknote classification;
  • Third layer According to different image quality of the same denomination, such as new banknotes, old banknotes, classify banknotes again;
  • Fourth layer Last According to the different orientation of the same denomination, such as: forward, reverse, re-classify the banknotes.
  • the classification discriminant function is defined as:
  • the input D can be classified as ⁇ class; otherwise, the input D does not belong to the category.
  • Image restoration for the localized interest (ROI) region of the old banknote that is, combining the information obtained by the fast classification process, first positioning the image's localized interest (ROI) region, and using the partial differential equation based image restoration technique to locate the interested In the (ROI) area image, the image of the location of interest (ROI) area is restored.
  • the specific implementation process is shown in Figure 5. After the image of the location of interest (ROI) region is restored, the image quality requirements for authentication are achieved.
  • the goal of image restoration is to recover the original image from a degraded image (noise, blur, and stain). While removing noise and blur, the boundary and detail information of the image can be preserved.
  • the recovery model needs to satisfy: (a) It should have an isotropic diffusion effect in a region with a small gradient, and (b) Only evolve along a gradient direction in a region with a large gradient value.
  • the embodiment adopts an image restoration model based on partial differential equations as follows:
  • Q e R is a bounded open set; for variable parameters, V is a gradient operator; ⁇ ( ⁇ ) is a monotonically decreasing function; it is a non-local averaging filter.
  • ⁇ ( ⁇ ) uses the boundary function in the anisotropic diffusion model, as shown in the following equation.
  • is the boundary threshold parameter
  • Step 4 Identification identification.
  • banknote information entering this link is divided into two types: banknotes of poor quality that have been processed by the image recovery, old banknotes of the quality, and banknotes of good quality that have not been image-recovered.
  • the banknote information is matched to determine the authenticity of the current banknote.
  • Step 5 output the result.
  • the above-mentioned method for identifying valuable documents using banknote identification as an example because the method first classifies the value documents, and then performs image restoration processing only on the old coins with poor image quality according to the classification result, thereby saving a large amount of image restoration processing time. Improves the real-time response speed of the banknote recognition system.
  • the system for realizing the value document identification method has a module architecture as shown in FIG. 6, and includes: an image acquisition module, configured to collect image data of the value document;
  • a storage module configured to store the foregoing image data and required standard model data
  • a data processing module performs feature region feature extraction, fast classification, and image restoration processing on the stored image data; that is, selecting a feature region of the value document, and extracting features of the value document from the feature region; Describe the characteristics of the value document, perform rapid classification processing on the input value document according to the pre-set classification model, select banknotes with good image quality and banknotes with poor image quality; and recover the image of banknotes with poor image quality, Obtaining the restored image;
  • An identification module configured to perform pseudo-identification processing on the processed image data
  • An output module configured to output an authentication result of the identification module
  • a central control module for controlling and coordinating the work of the above modules, including data transmission, resource allocation, and the like.
  • a data link is formed between the modules, and the data link is a well-known technology, and the present invention will not be described again.
  • the implementation process of the functions and functions of the modules in the system refer to the corresponding implementation process in the above method, and details are not described herein.
  • the present invention further provides a value document identification device, including: a selection module, an extraction module, a classification processing module, a recovery processing module, an identification module, and an output module, wherein the selection module is used for Selecting a feature area of the value document; the extracting module is configured to extract a feature of the value document from the feature area; the classification processing module is configured to input the image according to the extracted feature of the value document The value document is quickly classified according to a pre-set classification model, and the banknote with good image quality and the banknote with poor image quality are selected; the recovery processing module recovers the image of the banknote with poor image quality, and the processed image is processed.
  • a value document identification device including: a selection module, an extraction module, a classification processing module, a recovery processing module, an identification module, and an output module, wherein the selection module is used for Selecting a feature area of the value document; the extracting module is configured to extract a feature of the value document from the feature area; the classification processing module is configured to input the image according to the extracted
  • the identification module is configured to perform the forensic identification of the banknotes with good image quality obtained by the classification processing module, and to perform false identification on the image recovered by the recovery processing module to determine the authenticity of the current banknote;
  • An output module configured to output a result of the identification module determining the current banknote.
  • the value file identification device may be integrated in the service terminal, or may be deployed independently. This embodiment is not limited.
  • the implementation process of the functions and functions of the modules in the device refer to the corresponding implementation process in the foregoing method, and details are not described herein again.
  • the system Since the system only performs image (ROI) region image restoration processing on the old banknote with poor image quality, and directly authenticates the new banknote with good image quality, it saves a lot of recognition processing time, so the real-time response speed is fast. It is suitable for financial self-service equipment, its low banknote rejection rate and easy to use, which is conducive to the promotion and use of financial self-service equipment.
  • ROI image
  • the present invention can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is a better implementation. the way.
  • the technical solution of the present invention may be embodied in the form of a software product in essence or in the form of a software product, which may be stored in a storage medium such as a ROM/RAM or a disk. , an optical disk, etc., includes instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) to perform the methods described in various embodiments of the present invention or portions of the embodiments.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Image Analysis (AREA)
  • Inspection Of Paper Currency And Valuable Securities (AREA)

Description

有价文件识别方法及其识别系统、 装置
本申请要求于 2010 年 8 月 3 日提交中国专利局、 申请号为 201010251473.0、 发明名称为"有价文件识别方法及其识别系统"的中国专利申 请的优先权, 其全部内容通过引用结合在本申请中。
技术领域
本发明涉及图像处理及模式识别技术,特别涉及有价文件, 例如纸币的识 别方法及其识别系统、 装置。
背景技术
随着社会经济的快速增长, 纸币的流通量越来越大,银行传统的手工处理 方式难以满足大量纸币处理的需求。 为了提高纸币处理效率, 降低人力成本, 急需大量易用性好、可靠性高的金融自助设备投放社会使用。但在流通的纸币 中, 旧钞的比例高, 绝大多数的旧钞图像也存在笔迹、 划痕、 污损、 褶皱等退 化情况, 增加了鉴伪识别的难度, 而目前国、 内外设备制造商生产的金融自助 设备对残钞和旧钞的适应性差, 导致已投放到市场的金融自助设备拒钞率高、 易用性差, 不利于金融自助设备的推广使用。如果把旧钞的图像质量恢复到新 钞的图像质量水平, 就有可能解决金融自助设备对旧钞的拒钞率高的问题。基 于偏微分方程的图像恢复技术, 目前取得一定的研究成果, 它在处理划痕、 笔 迹、 污损、 褶皱等退还现象, 能够取得较好的恢复效果, 旧钞图像的恢复与基 于偏微分方程的图像恢复的研究内容是一致的。基于偏微分方程的图像恢复技 术应用于钞票图像恢复, 可以取得较好的恢复效果。
图 1为传统的钞票图像识别处理流程图,如图所示,先对整幅图像做图像 恢复处理, 再提取特征, 然后按照币种、 面额或新旧等标准分类, 最后才是鉴 伪识别。
令 T表示每张钞票识别时间, 令^。表示图像处理的时间开销, ^表示特征 提取的时间开销, 表示分类的时间开销, ^表示鉴伪的时间开销, 令7^表 示系统的实时响应的时间范围, 则当且仅当识别系统满足(1 ) 式时, 才能达 到实际应用的需求: T = ta + tb + tc "d ( 1 )
\ _t ,
― max
对于带有笔迹、 划痕、 污损、 褶皱等退化情况的旧钞票, 由于钞票退化会 降低系统鉴伪的精度, 因此, 系统需要先对纸币的 "整幅图像"作图像恢复处 理,需要大量的时间开销,会出现图像处理时间 ^。远大于分类时间 ,即 》 。
然而, 钞票识别系统应用于金融自助设备时, 要求具备较高的实时性:每 张钞票的识别必须在有限的时间内完成。现有技术中的钞票识别系统,在识别 过程中先对纸币的 "整幅图像"进行图像恢复处理, 占用了大量的时间; 另夕卜, 现有技术对图像质量好的纸币也做恢复处理, 浪费了系统资源。每张钞票的识 别时间开销剧增, 钞票识别系统难以满足应用的实时性需求7 具体情况如 (2)式所示:
Figure imgf000004_0001
因此, 已有的钞票识别系统为了提高实时响应速度, 对旧钞不作处理, 即 拒绝识别旧钞, 从而导致大量旧钞被拒收, 钞票识别系统的拒钞率高, 影响了 钞票识别系统的易用性, 不利于金融自助设备的推广应用。
发明内容
本发明的目的在于提供一种实时响应速度快的有价文件识别方法,以及实 时响应速度快、 拒钞率低的有价文件识别系统、 装置。
该有价文件识别方法包括:
步骤 1, 提取特征, 选取有价文件的特征区域, 并从所述特征区域中提取 用于快速分类的有价文件特征;
步骤 2, 快速分类, 根据步骤 1中提取的有价文件特征, 对输入的有价文 件根据预先设置的分类模型做快速分类处理, 获得有价文件的币种、 面额、 朝 向以及图像质量信息等信息,并遴选出图像质量好的纸币一一新钞和图像质量 差的纸币——旧钞;
步骤 3, 图像恢复, 应用基于偏微分方程的图像恢复技术对旧钞图像进行 恢复处理, 得到恢复处理后的图像; 步骤 4, 鉴伪处理, 对新钞直接做鉴伪识别; 对旧钞则通过恢复后的图像 做鉴伪识别, 判定当前钞票的真实性;
步骤 5 , 输出判定当前钞票的结果。
进一步的, 步骤 1中所述提取特征是结合钞票的多波长图像特性、钞票图 像的多分辨率特性、 以及钞票图像的防伪属性,选取钞票多波长图像的特征区 域, 结合分类模型, 从选定的特征区域中提取特征。
进一步, 步骤 1中所述有价文件特征包括白水印、 黑水印、 变色油墨、 磁 性安全线、 缩微文字、 凹版印刷图案、 面额数字号码以及钞票冠字号码中至少 一种。
进一步的, 步骤 1中所述有价文件特征的向量化, 具体包括: 对有价文件 的特征区域图像作归一化处理, 得到归一化图像; 从所述归一化图像中选取" 个不同的特征区域,对于 1至 n个特征区域图像分别进行灰度值平均计算; 根 据上述 1至 n个灰度平均值形成特征向量。 具体表示如下:
假设对于高为 宽为 W的钞票灰度图像/ ,) , 其中, ^表示行号 ( ^ (1, )), y表示列号( ), 表示在坐标为(U)的像素灰度值, 首先对特征区域图像作归一化处理, 得到归一化图像 , ; 再从钞票图像中 选取"个不同的特征区域, 对于高为^' , 宽为 W '的特征区域图像 ^ , 其中 = (1,2,···,Μ , 选取其灰度平均值作为它们的特征值, 即
Μ Ν d = x=l y=l
1 M x N
最后, 按上式计算 , 得到特征向量^^ ^ ^^
进一步的, 所述快速分类模型是根据不同的币种、 面额、 朝向以及图像质 量建立的。
进一步的, 所述快速分类模型包括如下层级:
第一层: 不同币种, 如: 人民币、 欧元、 美元;
第二层: 同一币种的不同面额;
第三层: 同一面额的不同朝向; 第四层: 同一面额的不同图像质量, 包括新钞和旧钞。
可选择的, 所述分类模型可以为如下层级顺序:
第一层: 不同币种, 如: 人民币、 欧元、 美元;
第二层: 同一币种的不同面额;
第三层: 同一面额的不同图像质量, 包括新妙和旧钞;
第四层: 同一面额的不同朝向。
进一步的, 分类模型通过如下步骤建立: 划分 p 个币种, 每种币种的 q 类面额, 每个面额的正反 2面朝向以及新、 旧图像的特征区域; 分别对上一步 形成的 Ρχ^χ4个不同图像作归一化处理, 得到归一化图像; 从每个归一化图 像中选取"个不同的特征区域, 对于 1至 η个特征区域图像分别进行灰度值平 均计算, 形成特征向量; 通过 R 个训练样本对所述特征向量进行训练形成 pxgx4个类聚中心, 依此建立有价文件特征区域的分类模型。 具体如下所示: 令^{^'",^}表示所有币种; 其中, -^,^…,^表示第 /类币 种包含 ^类面额; 令^^ ,0^表示钞票的朝向, 表示正向, "2表示反向;
B = {A,A表示图像质量, A表示新, A表示旧; };
则可得 px^4个类别,每种类 C = l,2"*,pxgx4)表示,如下式所述:
Figure imgf000006_0001
用 表示类别 的特征向量, 则有
D = {D1,D2,^;Dpxqx4} 对于上述每个类别 ^( = 12,''',Ρ>^Χ4), 选取 R个训练样本, 则类别 ^ ^的 聚类中心可用下式描述:
其中, _ "。 进一步的, 快速分类处理包括如下步骤: 对于提取的特征, 与每个类别的聚类中心 的距离为:
ek =||Ζ)-Ζ)Α||(^ = 1,2,···,ρχ^χ4) 分类判别函数定义为:
g ,· (D) = min( k ) = minllZ) - Dk \\(k = 1,2,· ·;ρχρχ4)
J k k " "
Figure imgf000007_0001
其中, 当 [0.1-0.5], 可把输入 D划归为 类; 否则, 输入 D不属于 类 别。
进一步的, 步骤 3中所述图像恢复, 其恢复对象为图像质量差的纸币一一 旧钞的感兴趣区域(R0I)。
进一步的,对旧钞图像进行恢复是采用基于偏微分方程的图像恢复方法完 成的。
本发明提供的有价文件识别系统包括:
一图像采集模块, 用于采集有价文件的图像数据;
一存储模块, 用于存储上述图像数据以及所需标准模型数据;
一数据处理模块,对上述存储图像数据进行特征区域特征提取、快速分类、 图像恢复处理; 即选取有价文件的特征区域, 并从所述特征区域中提取有价文 件的特征; 根据提取的所述有价文件的特征,对输入的有价文件按照预先设置 的分类模型做快速分类处理, 遴选出图像质量好的纸币和图像质量差的纸币; 对图像质量差的纸币的图像进行恢复处理,得到恢复处理后的图像(即图像数 据);
一识别模块, 用于对上述处理后的图像数据与标准模型数据比对, 进行鉴 伪识别处理;
一输出模块, 用于将识别模块的鉴伪识别结果输出;
一中央控制模块,用于控制和协调上述模块(即图像采集模块、存储模块、 数据处理模块、 识别模块和输出模块)之间的工作, 包括数据传输、 资源分配 等。
本有价文件识别系统中的各模块之间形成数据链路,其中数据链路属于公 知技术, 本发明不再赘述。
本发明提供的一种有价文件识别装置, 包括:
选取模块, 用于选取有价文件的特征区域;
提取模块, 用于从所述特征区域中提取有价文件的特征;
分类处理模块, 用于根据提取的所述有价文件的特征,对输入的有价文件 按照预先设置的分类模型做快速分类处理,遴选出图像质量好的纸币和图像质 量差的纸币;
恢复处理模块,对图像质量差的纸币的图像进行恢复处理,得到处理后的 图像;
识别模块, 用于分类处理模块得到的图像质量好的纸币直接做鉴伪识别; 以及对恢复处理模块恢复处理后的图像做鉴伪识别, 判定当前纸币的真实性; 以及
输出模块, 用于输出识另模块判定当前纸币的结果。
该有价文件识别方法、装置及系统的有益效果是: 省略了对图像质量好的 纸币一一新钞的图像恢复处理, 节省了时间, 提高了整个系统的处理效率; 且 对图像质量差的纸币——旧钞的感兴趣区域(ROI )做恢复处理, 既节约系统 开销,又为鉴伪环节提供质量好的图像数据,降低鉴伪难度,提高鉴伪的精度, 从而提高了钞票识别系统的接收率。
附图说明
图 1是传统的钞票图像识别处理方法的流程图;
图 2是本发明具体实施例提供的有价文件识别方法的流程图;
图 3是图 2中建立快速分类模型的示意图;
图 4是图 2中建立快速分类的另一种示意图;
图 5是图 2中图像恢复的示意图; 以及
图 6是本发明具体实施例提供的有价文件识别系统的结构示意图。
具体实施方式
为了使本技术领域的人员更好地理解本发明实施例的方案,下面结合附图 和实施方式对本发明实施例作进一步的详细说明。 以下结合图示举例说明本发明提供的有价文件识别方法的流程步骤以及 有价文件识别系统的模块框架。
如图 2所示, 该有价文件识别方法包括: 步骤 1 , 提取特征; 步骤 2, 快 速分类; 步骤 3, 图像恢复; 步骤 4, 鉴伪识别; 以及步骤 5, 输出结果。 以 下详细介绍各步骤的具体内容。
步骤 1 , 提取特征。
结合钞票的多波长图像特性、钞票图像的多分辨率特性、 以及钞票图像的 防伪属性, 选取钞票多波长图像的特征区域, 结合分类模型, 从选定的特征区 域中提取特征, 如: 白水印、 黑水印、 变色油墨、 磁性安全线、 缩微文字、 凹 版印刷图案、 面额数字号码和 /或钞票冠字号码等。
特征的量^ ^表示如下:
假设对于高为 宽为 W的钞票灰度图像/ , 其中, 表示行号 ( (1, H) ), y表示列号( y e (i,w) ), 表示在坐标为(U)的像素灰度值。
首选, 对特征区域图像作归一化处理, 得到归一化图像 :
其次, 从钞票图像中选取"个不同的特征区域, 对于高为^' , 宽为 ^的 特征区域图像 ^, , 其中'' = (1,2,*^,"), 选取其灰度平均值作为该钞票图像 的特征值, 即
1 Μ χ Ν
最后, 按上式计算 , 得到特征向量^^ ^^,^^
步骤 2, 快速分类。
对输入的钞票信息根据预先设置的分类模型做快速分类处理,得到钞票的 币种、 面额、 朝向和图像质量。 其中, 币种、 面额、 朝向合图像质量为后续环 节定位感兴趣(ROI ) 区域提供指导法则。 例如: 人民币 100元正面, 有白水 印、 光变油墨面额数字以及毛主席头像; 人民币 100元反面, 有隐形文字、 有 色荧光红色条纹以及凹版印刷面额数字。
图像质量则决定是否需要做图像恢复处理,图像质量好的纸币直接送识别 处理, 图像质量差的旧钞送数据处理模块处理。
其中, 分类模型的建立的示意图如图 3所示,
首先, 根据不同的币种、 面额和图像质量以如下几个层级建立分类模型: 第一层: 根据不同币种, 如: 人民币、 欧元、 美元, 将钞票进行分类; 第二层: 根据同一币种的不同面额, 如: 人民币 100元、 50元, 对钞票 做进一步的分类;
第三层: 再根据同一面额的不同朝向, 如: 正向、 反向, 对钞票进行第三 层次的分类; 然后
第四层: 根据同一面额的不同图像质量, 如: 新钞、 旧钞, 对钞票再次分 类。
其中第三层和第四层可以互换顺序,如图 4所示,第一层:根据不同币种, 如: 人民币、 欧元、 美元, 对钞票进行分类; 第二层: 根据同一币种的不同面 额, 如: 人民币 100元、 50元, 对钞票进行第二层次的分类; 第三层: 根据 同一面额的不同图像质量, 如新钞、 旧钞, 再次对钞票分类; 第四层: 最后根 据同一面额的不同朝向, 如: 正向、 反向, 对钞票做再次分类。
其次, 建立聚类中心
令^ ^,…, 表示所有币种; 其中, ψ'= , 表示第 '·类币 种包含 ^类面额;
令 ^ = {«"«2}表示钞票的朝向, 表示正向, 《2表示反向;
B = {A,A表示图像质量, A表示新, A表示旧; }
则可得/^ <4个类别,每种类别用<^ = 1,2"*,/^^4)表示,如下式所述: C,
Figure imgf000010_0001
用 表示类别 的特征向量, 则有
D = {D1,D2,^;Dpxqy 对于上述每个类别 ^^12,"',^^4), 选取 R个训练样本, 则类别 的 聚类中心可用下式描述:
Dk = {¾1, 2,· · ·, „} 其中, = R 依据上述分类模型进行分类:
对于提取的特征 与每个类别的聚类中心
Figure imgf000011_0001
分类判别函数定义为:
g · (D) = min( k ) = minllZ) - Dk \\(k = 1,2,·•;pxqx4)
其中, 当 ζ [0.1-0.5]时, 可把输入 D划归为 ^类; 否则, 输入 D不属于 类别。
步骤 3, 图像恢复。
针对旧钞的定位感兴趣(ROI) 区域做图像恢复, 即结合快速分类环节获 得的信息, 先定位图像的定位感兴趣(ROI) 区域, 把基于偏微分方程的图像 恢复技术用到定位感兴趣(ROI) 区域图像中, 对定位感兴趣(ROI) 区域图 像做恢复处理。 具体实现流程如图 5所示, 定位感兴趣(ROI) 区域图像经恢 复处理后, 达到鉴伪的图像质量要求。
图像恢复的目标是从一幅降质图像(噪声、 模糊、 污损 )恢复出原始的图 像, 在去除噪声和模糊的同时, 图像的边界和细节信息能 4艮好地保留下来。 为 达到上述条件, 恢复模型需要满足: (a)在梯度较小的区域应具有各向同性的 扩散效应, ( b )在梯度值较大的区域只沿着梯度方向演化。 为此, 本实施例采 用基于偏微分方程的图像恢复模型如下:
Figure imgf000011_0002
其中 是实时获得的观测图像, 是原始图像; 利用梯度下降法, 引入时 间变量 图像 可以被看作是关于时间变化的函数, 设退化图像为初始时刻 的时间函数, 即/ 0^ = 0) = /, Λ=/0^,Ο。
Qe R "是一个有界开集; 为可变参数, V为梯度算子; φ(·)是一个单调递减函数; 是非局部的均值滤波器。
(1) φ(·)采用 ΡΜ异向扩散模型中的边界函数, 如下式所示。
Φ(s) = l + ^ - ^ (5)
1+(— )2
Μ
其中, Μ为边界阈值参数。
(2)非局部的均值滤波器 如下式所示: φ( x) =
Figure imgf000012_0001
式(4) 的梯度下降方程为: ft =f -fo + Μν(Φ( (/0)) I V/ ^-1 |^|) (7) 其中, (·)表示散度。
为避免分母为零, 给分母加一个无穷小, 把上式变为: ft =f -fo+ Μν(Φ( ( 0 )) I V/ 1 (/。))- 1 , V ) (8) 以实时观测到的定位感兴趣( ROI ) 区域图像为 ; 标准 ROI图像为 ,。。 采用高斯-塞德尔法对式( 8 )进行迭代, 完成 ROI图像 的恢复处理。
步骤 4, 鉴伪识别。
进入该环节的钞票信息分为两种:图像恢复处理过的质量差的纸币一一旧 钞, 以及没做图像恢复处理的质量好的纸币一一新钞。
根据事先存储的纸币标准模型, 与纸币信息进行匹配, 判定当前钞票的真 实性。
步骤 5, 输出结果。
将货币真实性结果输出。
上述以钞票识别为例的有价文件识别方法, 由于该方法先对有价文件分 类, 然后根据分类结果仅对图像质量差的旧币做图像恢复处理, 因此节约了大 量的图像恢复处理时间, 提高了钞票识别系统的实时响应速度。 实现该有价文件识别方法的系统, 其模块架构如图 6所示, 包括: 一图像采集模块, 用于采集有价文件的图像数据;
一存储模块, 用于存储上述图像数据以及所需标准模型数据;
一数据处理模块,对上述存储图像数据进行特征区域特征提取、快速分类、 图像恢复处理; 即选取有价文件的特征区域, 并从所述特征区域中提取有价文 件的特征; 根据提取的所述有价文件的特征, 对输入的有价文件按照预先设置 的分类模型做快速分类处理, 遴选出图像质量好的纸币和图像质量差的纸币; 对图像质量差的纸币的图像进行恢复处理, 得到恢复处理后的图像;
一识别模块, 用于对上述处理后的图像数据进行鉴伪识别处理;
一输出模块, 用于将识别模块的鉴伪识别结果输出; 以及
一中央控制模块, 用于控制和协调上述模块的工作, 包括数据传输、 资源 分配等。
各模块之间形成数据链路,其中数据链路属于公知技术,本发明不再赘述。 所述系统中各个模块的功能和作用的实现过程详见上述方法中对应的实 现过程, 在此不再赘述。
基于上述方法的实现过程, 本发明还提供一种有价文件识别装置, 包括: 选取模块, 提取模块, 分类处理模块, 恢复处理模块, 识别模块和输出模块, 其中, 所述选取模块, 用于选取有价文件的特征区域; 所述提取模块, 用于从 所述特征区域中提取有价文件的特征; 所述分类处理模块, 用于根据提取的所 述有价文件的特征,对输入的有价文件按照预先设置的分类模型做快速分类处 理, 遴选出图像质量好的纸币和图像质量差的纸币; 所述恢复处理模块, 对图 像质量差的纸币的图像进行恢复处理, 得到处理后的图像; 所述识别模块, 用 于分类处理模块得到的图像质量好的纸币直接做鉴伪识别;以及对恢复处理模 块恢复处理后的图像做鉴伪识别,判定当前纸币的真实性;以及所述输出模块, 用于输出识别模块判定当前纸币的结果。
其中, 所述有价文件识别装置可以集成在服务终端中, 也可以独立部署, 本实施例不作限制。 所述装置中各个模块的功能和作用的实现过程详见上述方法中对应的实 现过程, 在此不再赘述。
由于该系统仅对图像质量差的旧钞进行感兴趣( ROI )区域图像恢复处理, 以及对图像质量好的新钞直接进行真伪鉴别, 因此节约了大量的识别处理时 间, 因此实时响应速度快,适合应用于金融自助设备,其拒钞率低, 易用性好, 利于金融自助设备的推广使用。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到本发明 可借助软件加必需的通用硬件平台的方式来实现, 当然也可以通过硬件,但很 多情况下前者是更佳的实施方式。基于这样的理解, 本发明的技术方案本质上 或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机 软件产品可以存储在存储介质中, 如 ROM/RAM、 磁碟、 光盘等, 包括若干指 令用以使得一台计算机设备(可以是个人计算机, 服务器, 或者网络设备等) 执行本发明各个实施例或者实施例的某些部分所述的方法。
以上所述仅是本发明的优选实施方式,应当指出,对于本技术领域的普通 技术人员来说, 在不脱离本发明原理的前提下, 还可以作出若干改进和润饰, 这些改进和润饰也应视为本发明的保护范围。
+

Claims

权 利 要 求
1、 一种有价文件识别方法, 其特征在于, 包括:
选取有价文件的特征区域, 并从所述特征区域中提取有价文件的特征; 根据提取的所述有价文件的特征,对输入的有价文件按照预先设置的分类 模型做快速分类处理, 遴选出图像质量好的纸币和图像质量差的纸币;
对图像质量差的纸币的图像进行恢复处理, 得到恢复处理后的图像; 对图像质量好的纸币直接做鉴伪识别; 并对恢复处理后的图像做鉴伪识 另1 J , 判定当前纸币的真实性; 以及
输出判定当前纸币的结果。
2、 如权利要求 1所述的有价文件识别方法, 其特征在于, 所述提取有价 文件的特征是结合纸币的多波长图像特性、纸币图像的多分辨率特性、 以及纸 币图像的防伪属性, 选取纸币多波长图像的特征区域, 结合分类模型, 从选定 的特征区域中提取特征。
3、 如权利要求 1所述的有价文件识别方法, 其特征在于, 所述有价文件 的特征包括白水印、 黑水印、 变色油墨、 磁性安全线、 缩微文字、 凹版印刷图 案、 面额数字号码和 /或钞票冠字号码。
4、 如权利要求 1所述的有价文件识别方法, 其特征在于, 还包括: 对提 取有价文件的特征进行向量化, 具体包括步骤:
对有价文件的特征区域图像作归一化处理, 得到归一化图像;
从所述归一化图像中选取 w个不同的特征区域, 对于 1至 n个特征区域图 像分别进行灰度值平均计算;
根据上述 1至 n个灰度平均值形成特征向量。
5、 如权利要求 4所述的有价文件识别方法, 其特征在于, 所述分类模型 是根据不同的币种、 面额、 朝向以及图像质量建立的。
6、 如权利要求 5所述的有价文件识别方法, 其特征在于, 还包括: 预先 设置分类模型, 具体包括如下步骤:
划分 p个币种, 每种币种的 q类面额, 每个面额的正反 2面朝向以及新、 旧图像的特征区域;
分别对上一步形成的 P X4个不同图像作归一化处理, 得到归一化图像; 从每个归一化图像中选取 w个不同的特征区域, 对于 1至 n个特征区域图 像分别进行灰度值平均计算, 形成特征向量;
通过 R个训练样本对所述特征向量进行训练形成 ΡΧ^Χ 4个类聚中心, 依 此建立有价文件特征区域的分类模型。
7、 如权利要求 6所述的有价文件识别方法, 其特征在于, 所述快速分类 处理是根据提取的有价文件的特征与分类模型之中每个类别的聚类中心的距 离进行分类。
8、 如权利要求 7所述的有价文件识别方法, 其特征在于, 当提取的有价 文件的特征与分类模型之中每个类别的聚类中心的距离属于 [0.1-0.5]之间时, 则将其归入该类别, 否则不属于该类别。
9、 如权利要求 1所述的有价文件识别方法, 其特征在于, 所述图像质量 差的纸币的图像进行恢复处理,其恢复对象为图像质量差的纸币的感兴趣区域 ( ROI )。
10、 如权利要求 1或 9所述的有价文件识别方法, 其特征在于, 所述图像 恢复是基于偏微分方程的图像恢复技术。
11、 一种有价文件识别系统, 其特征在于, 包括:
一图像采集模块, 用于采集有价文件的图像数据;
一存储模块, 用于存储所述图像数据以及所需标准模型数据;
一数据处理模块, 用于对存储所述图像数据进行特征区域特征提取、快速 分类、 图像恢复处理;
一识别模块, 用于对上述处理后的图像数据与标准模型数据比对, 进行鉴 伪识别处理;
一输出模块, 用于将识别模块的鉴伪识别结果输出; 以及
一中央控制模块, 用于控制和协调所述模块之间的数据传输。
12、 一种有价文件识别装置, 其特征在于, 包括: 选取模块, 用于选取有价文件的特征区域;
提取模块, 用于从所述特征区域中提取有价文件的特征;
分类处理模块, 用于根据提取的所述有价文件的特征,对输入的有价文件 按照预先设置的分类模型做快速分类处理,遴选出图像质量好的纸币和图像质 量差的纸币;
恢复处理模块,对图像质量差的纸币的图像进行恢复处理,得到处理后的 图像;
识别模块, 用于分类处理模块得到的图像质量好的纸币直接做鉴伪识别; 以及对恢复处理模块恢复处理后的图像做鉴伪识别, 判定当前纸币的真实性; 以及
输出模块, 用于输出识另模块判定当前纸币的结果。
PCT/CN2011/076550 2010-08-03 2011-06-29 有价文件识别方法及其识别系统、装置 WO2012016484A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
AU2011288069A AU2011288069B2 (en) 2010-08-03 2011-06-29 Valuable file identification method and identification system, device thereof
US13/810,422 US9262677B2 (en) 2010-08-03 2011-06-29 Valuable file identification method and identification system, device thereof
EP11814067.2A EP2602771A4 (en) 2010-08-03 2011-06-29 IDENTIFICATION METHOD AND VALUE DOCUMENT IDENTIFICATION SYSTEM AND CORRESPONDING DEVICE

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN2010102514730A CN101908241B (zh) 2010-08-03 2010-08-03 有价文件识别方法及其识别系统
CN201010251473.0 2010-08-03

Publications (1)

Publication Number Publication Date
WO2012016484A1 true WO2012016484A1 (zh) 2012-02-09

Family

ID=43263692

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2011/076550 WO2012016484A1 (zh) 2010-08-03 2011-06-29 有价文件识别方法及其识别系统、装置

Country Status (5)

Country Link
US (1) US9262677B2 (zh)
EP (1) EP2602771A4 (zh)
CN (1) CN101908241B (zh)
AU (1) AU2011288069B2 (zh)
WO (1) WO2012016484A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2945101A4 (en) * 2013-01-11 2016-09-28 Grg Banking Equipment Co Ltd METHOD AND RECOGNITION DEVICE FOR SHEET TYPE SUPPORT

Families Citing this family (37)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101908241B (zh) * 2010-08-03 2012-05-16 广州广电运通金融电子股份有限公司 有价文件识别方法及其识别系统
CN102289857B (zh) * 2011-05-19 2013-09-25 广州广电运通金融电子股份有限公司 有价文件识别方法及系统
CN102169608B (zh) * 2011-05-27 2013-03-13 徐端全 多光谱纸币验钞机对纸币的处理方法
CN102324134A (zh) * 2011-09-19 2012-01-18 广州广电运通金融电子股份有限公司 有价文件鉴别方法及其鉴别装置
US9367888B2 (en) * 2012-01-20 2016-06-14 Hewlett-Packard Development Company, L.P. Feature resolutions sensitivity for counterfeit determinations
CN102682514B (zh) 2012-05-17 2014-07-02 广州广电运通金融电子股份有限公司 一种纸类识别方法及相关装置
CN102890841B (zh) * 2012-10-08 2013-11-27 广州广电运通金融电子股份有限公司 一种有价文件鉴别方法及装置
CN103208148B (zh) * 2013-02-06 2014-12-10 深圳宝嘉电子设备有限公司 一种货币核查系统及其方法
CN103177128A (zh) * 2013-04-19 2013-06-26 中国建设银行股份有限公司 钞票冠字号信息的处理方法和系统
US9978113B2 (en) 2014-03-26 2018-05-22 Hewlett-Packard Development Company, L.P. Feature resolutions sensitivity for counterfeit determinations
US9299225B2 (en) * 2014-06-23 2016-03-29 Ncr Corporation Value media dispenser recognition systems
CN104346858B (zh) * 2014-10-10 2016-11-02 新达通科技股份有限公司 一种基于磁图像的纸币面值识别方法与装置
CN104537756B (zh) * 2015-01-22 2018-04-20 广州广电运通金融电子股份有限公司 一种基于Lab色彩空间的钞票分类鉴别方法和装置
CN104574642B (zh) * 2015-01-30 2017-04-26 广州广电运通金融电子股份有限公司 一种基于Lab色彩空间的人民币变色油墨检测方法和装置
CN104809801A (zh) * 2015-04-17 2015-07-29 深圳怡化电脑股份有限公司 一种纸币图像的处理方法
CN104867225B (zh) * 2015-05-15 2018-12-11 深圳怡化电脑股份有限公司 一种纸币的面向识别方法及装置
CN105184950A (zh) * 2015-06-03 2015-12-23 深圳怡化电脑股份有限公司 一种分析纸币新旧的方法及装置
CN105184953B (zh) * 2015-08-03 2017-12-08 深圳怡化电脑股份有限公司 一种纸币处理的方法及装置
CN105160756A (zh) * 2015-08-18 2015-12-16 深圳怡化电脑股份有限公司 一种纸币面向识别方法与装置
CN105354914B (zh) * 2015-08-18 2018-01-02 深圳怡化电脑股份有限公司 一种纸币折角面向识别方法和装置
CN105184955B (zh) * 2015-08-18 2018-03-02 深圳怡化电脑股份有限公司 一种纸币的识别方法和装置
WO2017079904A1 (zh) * 2015-11-10 2017-05-18 深圳怡化电脑股份有限公司 一种货币面值的识别方法及装置
CN105528825B (zh) 2015-12-02 2018-08-31 广州广电运通金融电子股份有限公司 有价文件自适应识别方法和装置
CN106447909B (zh) * 2016-09-23 2019-04-26 深圳怡化电脑股份有限公司 竖列冠字号识别方法及装置
CN106683263B (zh) * 2017-01-10 2019-07-19 中国人民银行印制科学技术研究所 有价票据的缺陷管理方法及系统
CN110770711B (zh) * 2017-03-31 2023-12-22 3M创新有限公司 基于图像的伪造品检测
CN107331026B (zh) * 2017-06-14 2019-12-10 深圳怡化电脑股份有限公司 一种纸币识别方法及装置
CN108268889B (zh) * 2017-12-28 2020-09-01 王道顺 对雕刻凹印真伪检测方法、检测平台及检测系统
CN108229460A (zh) * 2018-01-09 2018-06-29 佛山科学技术学院 一种水印图像ocr识别方法及其系统
CN108257292A (zh) * 2018-02-06 2018-07-06 深圳怡化电脑股份有限公司 纸币分类回收方法、装置和系统以及存储介质
CN108389309B (zh) * 2018-02-06 2021-01-01 深圳怡化电脑股份有限公司 一种识别纸币真伪的方法和系统
CN108492447B (zh) * 2018-03-20 2020-09-15 深圳怡化电脑股份有限公司 美元面向识别方法、电子装置及计算机可读存储介质
CN109389615B (zh) * 2018-09-29 2021-05-28 佳都科技集团股份有限公司 基于深度学习卷积神经网络的硬币识别方法及处理终端
CN109859373A (zh) * 2018-12-15 2019-06-07 深圳壹账通智能科技有限公司 基于图像识别的纸币面值计算方法、装置及相关设备
CN111667002B (zh) * 2020-06-05 2023-11-24 中国银行股份有限公司 一种币种识别方法、识别装置、以及电子设备
CN112200966A (zh) * 2020-09-28 2021-01-08 武汉科技大学 一种人民币纸币形成方式的鉴定方法
CN114025104B (zh) * 2021-11-03 2024-02-20 江苏金视传奇科技有限公司 一种彩色图像显示装置及彩色图像显示系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1139437C (zh) * 2001-11-28 2004-02-25 王晓光 一种人民币纸币清分机
WO2009096535A1 (ja) * 2008-01-31 2009-08-06 Aruze Corp. 紙葉類識別装置、及び紙葉類識別方法
CN101517619A (zh) * 2006-09-29 2009-08-26 阿鲁策株式会社 纸张识别装置
CN201383174Y (zh) * 2009-03-23 2010-01-13 宋白桦 一种纸币清分机
CN101908241A (zh) * 2010-08-03 2010-12-08 广州广电运通金融电子股份有限公司 有价文件识别方法及其识别系统

Family Cites Families (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5550189A (en) * 1992-04-17 1996-08-27 Kimberly-Clark Corporation Modified polysaccharides having improved absorbent properties and process for the preparation thereof
US5285196A (en) * 1992-10-15 1994-02-08 Texas Instruments Incorporated Bistable DMD addressing method
US7187795B2 (en) * 2001-09-27 2007-03-06 Cummins-Allison Corp. Document processing system using full image scanning
US6556689B1 (en) * 1998-05-29 2003-04-29 University Of Delaware Watermarking methods for digital images and videos
US6324298B1 (en) * 1998-07-15 2001-11-27 August Technology Corp. Automated wafer defect inspection system and a process of performing such inspection
US7269279B2 (en) * 2002-03-25 2007-09-11 Cummins-Allison Corp. Currency bill and coin processing system
US7256874B2 (en) * 2002-10-18 2007-08-14 Cummins-Allison Corp. Multi-wavelength currency authentication system and method
US7529422B2 (en) * 2004-09-22 2009-05-05 Siemens Medical Solutions Usa, Inc. Gradient-based image restoration and enhancement
JP4475143B2 (ja) * 2004-11-29 2010-06-09 セイコーエプソン株式会社 画像補正装置
US20070140551A1 (en) * 2005-12-16 2007-06-21 Chao He Banknote validation
US7657089B2 (en) * 2006-02-21 2010-02-02 Microsoft Corporation Automatic classification of photographs and graphics
US8096584B2 (en) * 2006-07-24 2012-01-17 3M Innovative Properties Company Document authentication using template matching with fast masked normalized cross-correlation
US8540158B2 (en) * 2007-12-12 2013-09-24 Yiwu Lei Document verification using dynamic document identification framework
FR2932588B1 (fr) * 2008-06-12 2010-12-03 Advanced Track & Trace Procede et dispositif de lecture d'une caracteristique physique sur un objet
DE102008041944A1 (de) 2008-09-10 2010-03-11 Bundesdruckerei Gmbh Datenverarbeitungsverfahren zur Generierung eines Klassifikators und zur Überprüfung der Echtheit eines Dokuments, Vorrichtung zur Überprüfung der Echtheit eines Dokuments und Computerprogrammprodukt

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1139437C (zh) * 2001-11-28 2004-02-25 王晓光 一种人民币纸币清分机
CN101517619A (zh) * 2006-09-29 2009-08-26 阿鲁策株式会社 纸张识别装置
WO2009096535A1 (ja) * 2008-01-31 2009-08-06 Aruze Corp. 紙葉類識別装置、及び紙葉類識別方法
CN201383174Y (zh) * 2009-03-23 2010-01-13 宋白桦 一种纸币清分机
CN101908241A (zh) * 2010-08-03 2010-12-08 广州广电运通金融电子股份有限公司 有价文件识别方法及其识别系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP2602771A4 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2945101A4 (en) * 2013-01-11 2016-09-28 Grg Banking Equipment Co Ltd METHOD AND RECOGNITION DEVICE FOR SHEET TYPE SUPPORT

Also Published As

Publication number Publication date
CN101908241A (zh) 2010-12-08
US20130121539A1 (en) 2013-05-16
EP2602771A4 (en) 2014-05-07
US9262677B2 (en) 2016-02-16
AU2011288069A1 (en) 2013-01-31
CN101908241B (zh) 2012-05-16
AU2011288069A9 (en) 2013-11-07
EP2602771A1 (en) 2013-06-12
AU2011288069B2 (en) 2014-02-13

Similar Documents

Publication Publication Date Title
WO2012016484A1 (zh) 有价文件识别方法及其识别系统、装置
CN104298989B (zh) 基于斑马线红外图像特征的鉴伪方法及其系统
Abburu et al. Currency recognition system using image processing
CN102722708B (zh) 一种薄片类介质分类方法及其装置
CN102800148A (zh) 一种人民币序列号识别方法
CN107437293A (zh) 一种基于票据全局特征的票据防伪鉴别方法
CN111275070B (zh) 一种基于局部特征匹配的签名验证方法及设备
Uddin et al. Image-based approach for the detection of counterfeit banknotes of Bangladesh
CN113688821A (zh) 一种基于深度学习的ocr文字识别方法
Rathee et al. Feature fusion for fake Indian currency detection
CN105654609A (zh) 纸币处理方法及系统
Dhar et al. Paper currency detection system based on combined SURF and LBP features
CN112200075A (zh) 一种基于异常检测的人脸防伪方法
Ponishjino et al. Bogus currency authorization using HSV techniques
Kumar et al. Study of shadow and reflection based image forgery detection
CN118135641B (zh) 基于局部伪造区域检测的人脸伪造检测方法
CN112215225A (zh) 一种基于计算机视觉技术的kyc证件核验方法
CN106846609A (zh) 一种基于感知哈希的纸币面值面向识别方法
KR101232684B1 (ko) 베이시안 접근법을 이용한 지폐 진위 감별 방법
Pujiputra et al. Ultraviolet rupiah currency image recognition using Gabor wavelet
CN110415424B (zh) 一种防伪鉴定方法、装置、计算机设备和存储介质
Naseem et al. Counterfeit Recognition of Pakistani Currency
Shiby et al. Detection of Fake Currency Using Image Processing
Kumar et al. Currency Authentication using Color Based Processing
Vishnu et al. Currency detection using similarity indices method

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11814067

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 13810422

Country of ref document: US

WWE Wipo information: entry into national phase

Ref document number: 2011814067

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2011288069

Country of ref document: AU

Date of ref document: 20110629

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE