CN110598589A - Image pyramid-based palm print identification method, system, device and medium - Google Patents

Abstract

The present invention relates to a palmprint recognition method, system, device and medium based on an image pyramid. The method includes acquiring an original palmprint image of a palmprint to be matched; performing preprocessing on each original template palmprint image and the original palmprint image respectively , to obtain the target template palmprint image set and the target palmprint image respectively; use the scale invariant feature transformation method to extract the target SIFT feature point set and the template feature point set of each target template palmprint image respectively; based on the image pyramid method, according to each target template palmprint image and each template feature point set and target palmprint image and target SIFT feature point set, calculate the offset between the target palmprint image and each target template palmprint image; In the fingerprint image set, the palmprint to be matched is identified and matched according to all offsets, and the identification result is obtained and output. The invention overcomes the influence of palmprint deformation, translation and noise on palmprint recognition, and obviously improves the palmprint recognition precision.

Description

A Palmprint Recognition Method, System, Device and Medium Based on Image Pyramid

technical field

The present invention relates to the technical field of biological feature identification, in particular to a palmprint identification method, system, device and medium based on an image pyramid.

Background technique

As an important aspect of biometric identification technology, palmprint recognition technology has great development potential and application prospects. Palmprint recognition technology has been developed for more than 20 years since it was researched in 1997. Due to its own advantages, such as: compared with iris recognition, it is easier to collect, has no harm to the human body, and the collection equipment is relatively cheap; Compared with face recognition, twins can be identified more easily; compared with fingerprint recognition, it has higher recognition accuracy, and the collection method can be non-contact collection; compared with handwriting and gait recognition, handwriting and gait are easy to be detected Imitation and biometric features are easy to change, while the stability of palmprint recognition is high. Therefore, palmprint recognition has gradually become one of the hot research directions in the field of biometric recognition, and related products are also gradually advancing.

Stable and reliable palmprint information, such as palmprint main line, palmprint folds, palmprint lines and other texture features, can achieve reliable identification. The texture feature structure of the palmprint is relatively unique, and this structure also makes the anti-noise ability of the algorithm processing process more prominent. Even if the image with poor quality or low resolution is directly input, the palmprint can be extracted relatively stably. main features of .

Compared with traditional biometric identification, palmprint recognition has many natural advantages in terms of universality, differentiation, invariance, easy collection, acceptability and hardware cost. These advantages are also a key point for the future development of palmprint recognition. Favorable conditions.

However, in the palmprint recognition technology, palmprint deformation, noise and palmprint blur are several key difficulties affecting the palmprint recognition effect in the process of recognizing low-resolution palmprints. One of the existing methods to solve the palmprint deformation is to extract the feature descriptors of the corresponding local areas of the two palmprints, and then combine the palmprint matching method based on coding to match and recognize, but this scheme ignores the palmprint translation and noise on the palmprint. Due to the influence of the fingerprint recognition method, the matching and recognition of the palmprint are still not accurate enough, the recognition accuracy is not high, and the recognition effect is not good.

Contents of the invention

The technical problem to be solved by the present invention is to provide a kind of palmprint recognition method, system, device and medium based on the image pyramid, which overcomes the impact of palmprint translation and noise on palmprint recognition, and significantly improves the deficiencies in the prior art. palmprint recognition accuracy.

The technical scheme that the present invention solves the problems of the technologies described above is as follows:

A kind of palmprint recognition method based on image pyramid, comprises the following steps:

Step 1: Obtain the original palmprint image of the palmprint to be matched;

Step 2: Preprocess each original template palmprint image in the preset original template palmprint image set to obtain the target template palmprint image set, and preprocess the original palmprint image to obtain the target palmprint image ;

Step 3: adopt the scale-invariant feature transformation method to extract respectively the target SIFT feature point set of the target palmprint image and the template feature point set corresponding to each target template palmprint image in the target template palmprint image set;

Step 4: Based on the image pyramid method, according to each target template palmprint image and the template feature point set corresponding to each target template palmprint image, and the target palmprint image and the target palmprint image The target SIFT feature point set corresponding to the image is calculated to obtain the offset between the target palmprint image and each target template palmprint image;

Step 5: In the palmprint image set of the target template, identify and match the palmprint to be matched according to all offsets, obtain and output the recognition result.

The beneficial effects of the present invention are: because the palmprint image is easily affected by noise, before extracting the target SIFT feature point set of the target palmprint image and the one-to-one corresponding template feature point set of each target template palmprint image, each The original template palmprint images are preprocessed respectively to obtain one-to-one corresponding target template palmprint images. These target template palmprint images constitute the target template palmprint image set, and the original palmprint images are also preprocessed to obtain the target Palmprint image, through preprocessing, can filter out the palmprint line features affected by noise, improve the noise resistance of the subsequent scale-invariant feature transformation method, effectively overcome the influence of noise on palmprint recognition, and at the same time strengthen the The contrast of each target template palmprint image and the texture feature of the target palmprint image is conducive to subsequent palmprint recognition and matching;

The scale-invariant feature transform method (Scale-invariant Feature Transform, SIFT method) is an algorithm for detecting local features. This method finds extreme points in the scale space, and extracts its position, scale and rotation invariants, extracts The extracted SIFT features remain unchanged to rotation, scaling, and brightness changes, and also maintain a certain degree of stability to viewing angle changes, affine transformations, and noise. Therefore, the target SIFT feature point set extracted by the present invention through the SIFT method and each The one-to-one corresponding template feature point set of the target template palmprint image further facilitates subsequent palmprint recognition and matching, overcomes the influence of palmprint translation on palmprint recognition, and significantly improves the recognition accuracy of palmprint;

The image pyramid is an effective but conceptually simple structure for explaining images with multiple resolutions. The pyramid of an image is exactly a series of image collections with pyramid-shaped arrangements that gradually reduce the resolution; the present invention extracts the target SIFT feature point set After the template feature point set corresponding to each target template palmprint image, the image pyramid method can be used to effectively correct the palmprint deformation in a local area, especially for the low-resolution palmprint recognition process. The influence of palmprint deformation on palmprint recognition is understood, and the method is flexible and simple, which is conducive to the subsequent calculation of the offset between the target palmprint image and each target template palmprint image, and further improves the palmprint recognition accuracy;

Since the offsets between the feature points of the same palmprint are small, and the offsets between the feature points of different palmprints are large, the target palmprint image and the target template palmprint can be judged by the offset. Which target template palmprint image in the image set is the best match, so as to identify the palmprint to be matched. The identification and matching method is relatively simple and effective, and the recognition accuracy of the palmprint is obviously improved.

On the basis of above-mentioned technical scheme, the present invention can also be improved as follows:

Further: in described step 2, the specific steps that obtain described target template palmprint image set include:

Step 2a.1: extract the region of interest of each original template palmprint image in the original template palmprint image set respectively, obtain the template palmprint ROI image set;

Step 2a.2: using the MFRAT filter method, filter each template palmprint ROI image in the template palmprint ROI image set respectively to obtain a template palmprint filter image set;

Step 2a.3: In each template palmprint filtering image of the template palmprint filtering image set, each pixel is encoded respectively to obtain the target template palmprint image set;

In said step 2, the specific steps of obtaining said target palmprint image include:

Step 2b.1: Extract the region of interest of the original palmprint image to obtain the palmprint ROI image;

Step 2b.2: using the MFRAT filtering method to filter the palmprint ROI image to obtain a palmprint filtered image;

Step 2b.3: Encoding each pixel in the palmprint filtered image to obtain the target palmprint image.

Further: the step 2a.2 specifically includes:

Step 2a.2.1: performing histogram equalization processing on each template palmprint ROI image in the template palmprint ROI image set respectively to obtain the first intermediate template palmprint image set;

Step 2a.2.2: performing normalization processing on each of the first intermediate template palmprint images in the first intermediate template palmprint image set to obtain a second intermediate template palmprint image set;

Step 2a.2.3: In each second intermediate template palmprint image of the second intermediate template palmprint image set, construct the first MFRAT filter function, with any pixel point as the first center point, establish p×p The first filter grid, and in the first filter grid, according to the first MFRAT filter function, calculate the first response value of the first center point in each direction, and according to each All the first response values in each direction obtain the first pixel cumulative value of the first central point in each direction;

The first MFRAT filter function of the first center point (x, y) in the θ _k direction is:

Wherein, (x, y) is the coordinate of the first central point in the second intermediate template palmprint image, and r (x, y) is the pixel value corresponding to the first central point (x, y), θ _k (k=1 ,2,…,6) are the selected six directions, which are 0, π/6, 2π/6, 3π/6, 4π/6 and 5π/6, is a straight line equation of the first filtering grid in the θ _k direction, is the first center point (x, y) in the direction of θ _k , the first response value of the line equation;

Step 2a.2.4: Traversing each pixel of the palmprint image of each second intermediate template, according to the method of step 2a.2.3, obtaining the first pixel cumulative value of each pixel in each direction respectively, and Obtain a corresponding template palmprint filter image corresponding to a second intermediate template palmprint image according to all first pixel cumulative values of all pixels;

Step 2a.2.5: Obtain the template palmprint filter image set according to all template palmprint filter images;

The step 2b.2 specifically includes:

Step 2b.2.1: performing histogram equalization processing on the palmprint ROI image to obtain the first intermediate palmprint image;

Step 2b.2.2: performing normalization processing on the first intermediate palmprint image to obtain a second intermediate palmprint image;

Step 2b.2.3: In the second intermediate palmprint image, construct the second MFRAT filter function, take any pixel as the second filter grid of p×p, and set up the second filter grid in the second filter grid Inside, according to the second MFRAT filter function, calculate and obtain the multiple second response values of the second central point in each direction respectively, and obtain the second center according to all the second response values in each direction The accumulated value of the second pixel of the point in each direction;

The second MFRAT filter function of the second central point (x′, y′) in the direction of θ _k is:

Wherein, (x', y') is the coordinate of the second center point in the second middle palmprint image, and r'(x', y') is the pixel value corresponding to the second center point (x', y'), is a straight line equation of the second filtering grid in the θ _k direction, is the second center point (x′,y′) in the direction of θ _k , the second response value of the line equation;

Step 2b.2.4: traverse each pixel of the second intermediate palmprint image, and obtain the second pixel cumulative value of each pixel in each direction according to the method of step 2b.2.3;

Step 2b.2.5: Obtain the palmprint filtered image according to the accumulated values of all second pixels of all pixels.

Further: the step 2a.3 specifically includes:

Step 2a.3.1: In each template palmprint filtering image in the template palmprint filtering image set, the direction corresponding to the maximum value of all the first pixel cumulative values of each pixel is taken as the first pixel of the corresponding pixel. feature code value;

Step 2a.3.2: According to all the first feature coding values corresponding to all pixels in each template palmprint filtering image, obtain the first feature coding value subset corresponding to each template palmprint filtering image;

Step 2a.3.3: Obtain the target template palmprint image set according to all first feature code value subsets corresponding to all template palmprint filtered images;

The step 2b.3 specifically includes:

Step 2b.3.1: In the palmprint filtered image, use the direction corresponding to the maximum value of all the second pixel cumulative values of each pixel as the second feature encoding value of the corresponding pixel;

Step 2b.3.2: According to all the second characteristic coding values corresponding to all pixels in the palmprint filtering image, obtain the second characteristic coding value subset corresponding to the palmprint filtering image;

Step 2b.3.3: Obtain the target palmprint image according to the second feature code value subset.

Further: in said step 3, the specific steps of extracting the set of template feature points corresponding to each target template palmprint image in said target template palmprint image set include:

Step 3a.1: Using bilinear interpolation method, expand the palmprint image of each target template respectively to obtain the expanded target template palmprint image set;

Step 3a.2: In each enlarged target template palmprint image in the expanded target template palmprint image set, use the scale-invariant feature transformation method to construct the first scale space, and detect Obtain the first extremum point set corresponding to each enlarged target template palmprint image in the first scale space;

Step 3a.3: Use Harris Comer detector to filter the first extremum point set of each enlarged target template palmprint image in the first scale space to obtain one-to-one correspondence of each target template palmprint image Template feature point set;

In said step 3, the specific steps of extracting the target SIFT feature point set of said target palmprint image include:

Step 3b.1: using bilinear interpolation method to expand the target palmprint image to obtain the enlarged target palmprint image;

Step 3b.2: In the enlarged target palmprint image set, a scale-invariant feature transformation method is used to construct a second scale space, and the enlarged target palmprint image is detected according to the preset second pixel threshold. The corresponding second set of extremum points in the second scale space;

Step 3b.3: Using a Harris Comer detector to filter all the second extreme points in the second extreme point set to obtain the target SIFT feature point set corresponding to the target palmprint image.

Further: the concrete steps of described step 4 include:

Step 4.1: based on the image pyramid method, according to each target template palmprint image and the corresponding template feature point set, obtain the one-to-one corresponding template corner point coordinate set of each target template palmprint image, according to the target palmprint image and The target SIFT feature point set obtains a set of target corner coordinates corresponding to the target palmprint image;

Step 4.2: Using the BLPOC method, according to each template corner point coordinate set and the target corner point coordinate set, respectively calculate the offset between the target palmprint image and each target template palmprint image.

Further: the concrete realization of described step 5 is:

Taking the target template palmprint image corresponding to the minimum value of all offsets as the recognition result of the palmprint to be matched and outputting it.

According to another aspect of the present invention, a kind of palmprint recognition system based on image pyramid is provided, including acquisition module, preprocessing module, extraction module, calculation module and identification module:

The acquisition module is used to acquire the original palmprint image of the palmprint to be matched;

The preprocessing module is used to preprocess each original template palmprint image in the preset original template palmprint image set to obtain the target template palmprint image set, and is also used to preprocess the original palmprint image. Process to obtain the target palmprint image;

The extraction module is used to extract the target SIFT feature point set of the target palmprint image and the one-to-one corresponding template feature points of each target template palmprint image in the target template palmprint image set by adopting the scale-invariant feature transformation method set;

The calculation module is used to calculate the offset information between the target SIFT feature point set and the one-to-one corresponding template feature point set of each target template palmprint image based on the image pyramid method;

The identification module is configured to identify and match the palmprint to be matched according to all offset information in the palmprint image set of the target template, obtain and output the identification result.

The beneficial effects of the present invention are: obtain the original palmprint image of the palmprint to be matched by the acquisition module, and then preprocess each original template palmprint image by the preprocessing module to obtain a one-to-one corresponding target template palmprint image, These target template palmprint images constitute the target template palmprint image set, and the original palmprint images are also preprocessed to obtain the target palmprint image. Through preprocessing, the palmprint line features affected by noise can be filtered out to improve The noise resistance of the subsequent scale-invariant feature transformation method is improved, and the influence of noise on palmprint recognition is effectively overcome. At the same time, the contrast between the texture features of each target template palmprint image and the target palmprint image is enhanced, which is beneficial to subsequent Palmprint recognition and matching; the target SIFT feature point set extracted by the SIFT method through the extraction module and the template feature point set corresponding to each target template palmprint image are further conducive to subsequent palmprint recognition and matching, and overcome The impact of palmprint translation on palmprint recognition is clearly improved, and the recognition accuracy of palmprint is significantly improved; after extracting the target SIFT feature point set and the template feature point set corresponding to each target template palmprint image, the calculation module adopts the image The pyramid method can effectively correct the palmprint deformation in a local area, especially for the low-resolution palmprint recognition process, which effectively overcomes the influence of palmprint deformation on palmprint recognition. The method is flexible and simple, which is conducive to subsequent calculation goals The offset between the palmprint image and each target template palmprint image further improves the recognition accuracy of the palmprint; finally, the recognition module can judge which one of the target palmprint image and the target template palmprint image set The target template palmprint image is the most matched, so as to identify the palmprint to be matched. The recognition and matching method is relatively simple and effective, and the recognition accuracy of the palmprint is obviously improved.

On the basis of above-mentioned technical scheme, the present invention can also be improved as follows:

Further: the preprocessing module includes a region of interest extracting unit, a filtering unit and a coding unit;

The region of interest extraction unit is used to extract the region of interest of each original template palmprint image in the original template palmprint image set respectively to obtain the template palmprint ROI image set; it is also used to extract the original palmprint image The region of interest of the palmprint ROI image is obtained;

Described filtering unit is used for adopting MFRAT filter method, each template palmprint ROI image of described template palmprint ROI image set is carried out filter processing respectively, obtains template palmprint filter image set; Also for adopting MFRAT filter method, for The palmprint ROI image is filtered to obtain a palmprint filtered image;

The encoding unit is used to encode each pixel in each template palmprint filter image of the template palmprint filter image set to obtain the target template palmprint image set; In the palmprint filtered image, each pixel is coded separately to obtain the target palmprint image.

Further: the filtering unit is specifically used for:

Carry out histogram equalization processing to each template palmprint ROI image in the template palmprint ROI image set respectively, obtain the first intermediate template palmprint image set;

Each of the first intermediate template palmprint images in the first intermediate template palmprint image set is respectively normalized to obtain the second intermediate template palmprint image set;

In each second intermediate template palmprint image of the second intermediate template palmprint image set, build the first MFRAT filter function, take any pixel point as the first center point, and set up the first filter net of p*p Grid, and in the first filter grid, according to the first MFRAT filter function, calculate the first response values of the first central point in each direction, and according to all the first response values in each direction The cumulative value of the first pixel in each direction to the first center point;

The first MFRAT filter function of the first center point (x, y) in the θ _k direction is:

Wherein, (x, y) is the coordinate of the first central point in the second intermediate template palmprint image, and r (x, y) is the pixel value corresponding to the first central point (x, y), θ _k (k=1 ,2,…,6) are the selected six directions, which are 0, π/6, 2π/6, 3π/6, 4π/6 and 5π/6, is a straight line equation of the first filtering grid in the θ _k direction, is the straight line equation of the first center point (x, y) in the direction of θ _k The first response value on ;

Traverse each pixel of each second intermediate template palmprint image, according to the method of step 2a.2.3, obtain the first pixel cumulative value of each pixel in each direction respectively, and according to all pixels All the first pixel cumulative values obtain a corresponding template palmprint filter image corresponding to a second intermediate template palmprint image;

Obtain the template palmprint filtering image set according to all template palmprint filtering images;

The filtering unit is also specifically used for:

Carry out histogram equalization processing to described palmprint ROI image, obtain the first intermediate palmprint image;

Carry out normalization processing to described first intermediate palmprint image, obtain the second intermediate palmprint image;

In the second intermediate palmprint image, construct the second MFRAT filter function, take any pixel point as the second center point, set up the second filtering grid of p×p, and in the second filtering grid, according to The second MFRAT filter function calculates multiple second response values of the second central point in each direction, and obtains the second central point in each direction according to all the second response values in each direction. Accumulated value of two pixels;

The second MFRAT filter function of the second center point (x', y') in the direction of θ _k is:

Wherein, (x', y') is the coordinates of the second central point pixel in the second middle palmprint image, and r'(x',y') is the pixel corresponding to the second central point (x', y') value, is a straight line equation of the second filtering grid in the θ _k direction, is the straight line equation of the second center point (x′,y′) in the direction of θ _k The second response value on ;

Traversing through each pixel of the second intermediate palmprint image, according to the method of step 2b.2.3, obtain the second pixel cumulative value of each pixel in each direction respectively;

The palmprint filtered image is obtained according to the accumulated values of all second pixels of all pixels.

Further: the encoding unit is specifically used for:

In each template palmprint filtering image in the template palmprint filtering image set, the direction corresponding to the maximum value in all the first pixel cumulative values of each pixel point is used as the first feature encoding value of the corresponding pixel point;

According to all first feature coding values corresponding to all pixels in each template palmprint filtering image, obtain the first feature coding value subset corresponding to each template palmprint filtering image one-to-one;

Obtain the target template palmprint image set according to all first feature code value subsets corresponding to all template palmprint filtered images;

The coding unit is also specifically used for:

In the palmprint filtered image, the direction corresponding to the maximum value in all the second pixel cumulative values of each pixel is used as the second feature encoding value of the corresponding pixel;

According to all the second feature coding values corresponding to all pixels in the palmprint filtering image, obtain the second feature coding value subset corresponding to the palmprint filtering image;

The target palmprint image is obtained according to the second feature code value subset.

Further: the extraction module is specifically used for:

Using a bilinear interpolation method, the palmprint image of each target template is expanded separately to obtain an enlarged target template palmprint image set;

In each enlarged target template palmprint image in the expanded target template palmprint image set, a scale-invariant feature transformation method is used to construct a first scale space, and each enlarged target is detected according to a preset first pixel threshold A set of first extremum points corresponding one-to-one to the template palmprint image in the first scale space;

Adopt Harris Comer detector, filter the first extremum point collection of each enlarged target template palmprint image in the first scale space, obtain the template feature point set corresponding to each target template palmprint image one-to-one;

The extraction module is also specifically used for:

Expanding the target palmprint image by using a bilinear interpolation method to obtain the enlarged target palmprint image;

In the enlarged target palmprint image set, a scale-invariant feature transformation method is used to construct a second scale space, and according to a preset second pixel threshold, it is detected that the enlarged target palmprint image is in the second scale space The corresponding second set of extremum points;

A Harris Comer detector is used to filter all the second extreme points in the second extreme point set to obtain the target SIFT feature point set corresponding to the target palmprint image.

Further: the calculation module is specifically used for:

Based on the image pyramid method, according to each target template palmprint image and the corresponding template feature point set, obtain the one-to-one corresponding template corner point coordinate set of each target template palmprint image, according to the target palmprint image and the target SIFT feature point set, obtains the set of target corner coordinates corresponding to the target palmprint image;

Using the BLPOC method, according to each template corner point coordinate set and the target corner point coordinate set, the offset between the target palmprint image and each target template palmprint image is calculated respectively.

Further: the identification module is specifically used for:

The palmprint image of the target template corresponding to the minimum value of all offsets is used as the recognition result of the palmprint to be matched and output.

According to another aspect of the present invention, a palmprint recognition device based on an image pyramid is provided, including a processor, a memory, and a computer program stored in the memory and operable on the processor, the computer program The steps in the image pyramid-based palmprint recognition method of the present invention are implemented during operation.

The beneficial effects of the present invention are: through the computer program stored in the memory, and run on the processor, realize the palmprint recognition of the present invention, based on the scale-invariant feature transformation method and the image pyramid, effectively correct the deformation of the palmprint, It overcomes the influence of palmprint deformation, palmprint translation and noise on palmprint recognition, significantly improves the recognition accuracy of palmprint, and the recognition effect is good.

According to another aspect of the present invention, a kind of computer storage medium is provided, and described computer storage medium comprises: At least one instruction, when described instruction is carried out, realize the method in a kind of palmprint recognition method based on image pyramid of the present invention step.

The beneficial effects of the present invention are: by executing the computer storage medium containing at least one instruction, the palmprint recognition of the present invention is realized, based on the scale-invariant feature transformation method and the image pyramid, the palmprint deformation is effectively corrected, and the palmprint deformation is overcome , Palmprint translation and the impact of noise on palmprint recognition, significantly improve the recognition accuracy of palmprint, and the recognition effect is good.

Description of drawings

Fig. 1 is the schematic flow sheet of the palmprint recognition method based on image pyramid in the embodiment of the present invention one;

Fig. 2 is the schematic flow chart that obtains target template palmprint image set in the embodiment of the present invention one;

Fig. 3 is the schematic flow chart that obtains target palmprint image in the embodiment of the present invention;

4 is a schematic flow diagram of obtaining a template feature point set in Embodiment 1 of the present invention;

5 is a schematic flow diagram of obtaining a target SIFT feature point set in Embodiment 1 of the present invention;

FIG. 6 is a schematic flow chart of calculating an offset in Embodiment 1 of the present invention;

Fig. 7 is the structural representation one of the palmprint recognition system based on image pyramid in the embodiment of the present invention two;

FIG. 8 is a second structural schematic diagram of a palmprint recognition system based on an image pyramid in Embodiment 2 of the present invention.

Detailed ways

The principles and features of the present invention are described below in conjunction with the accompanying drawings, and the examples given are only used to explain the present invention, and are not intended to limit the scope of the present invention.

The present invention will be described below in conjunction with the accompanying drawings.

Embodiment one, as shown in Figure 1, a kind of palmprint recognition method based on image pyramid comprises the following steps:

S1: Obtain the original palmprint image of the palmprint to be matched;

S2: Preprocessing each original template palmprint image in the preset original template palmprint image set to obtain a target template palmprint image set, and performing preprocessing on the original palmprint image to obtain a target palmprint image;

S3: Using a scale-invariant feature transformation method to extract the target SIFT feature point set of the target palmprint image and the template feature point set corresponding to each target template palmprint image in the target template palmprint image set;

S4: Based on the image pyramid method, according to each target template palmprint image and the template feature point set corresponding to each target template palmprint image, and the target palmprint image and the target palmprint image The corresponding target SIFT feature point set is calculated to obtain the offset between the target palmprint image and each target template palmprint image;

S5: In the palmprint image set of the target template, perform recognition and matching on the palmprint to be matched according to all offsets, obtain and output a recognition result.

First, each original template palmprint image is preprocessed separately to obtain a one-to-one corresponding target template palmprint image. These target template palmprint images constitute the target template palmprint image set, and the original palmprint image is also preprocessed , to obtain the target palmprint image, through preprocessing, the palmprint line features affected by noise can be filtered out, the noise resistance of the subsequent scale-invariant feature transformation method is improved, and the influence of noise on palmprint recognition is effectively overcome. It also strengthens the contrast between the texture features of each target template palmprint image and the target palmprint image, which is conducive to subsequent palmprint recognition and matching; the target SIFT feature point set extracted by the SIFT method and the palmprint image of each target template The one-to-one corresponding template feature point set is further beneficial to the subsequent palmprint recognition and matching, overcomes the influence of palmprint translation on palmprint recognition, and significantly improves the recognition accuracy of palmprint; when extracting the target SIFT feature point set and each After one-to-one corresponding template feature point sets of target template palmprint images, the image pyramid method can be used to effectively correct the palmprint deformation in the local area. The method is flexible and simple, which is conducive to the subsequent calculation of target palmprint images and each The offset between the template palmprint images effectively overcomes the influence of palmprint deformation on palmprint recognition and further improves the recognition accuracy of palmprints; since the offsets between feature points of the same palmprint are small, and The offset between the feature points of different palmprints is relatively large, so it can be judged which target template palmprint image matches the target palmprint image and the target template palmprint image set by the offset, so as to identify the Matching palmprints, the recognition and matching method is relatively simple and effective, which obviously improves the recognition accuracy of palmprints.

It should be understood that among the present invention, the original template palmprint image set contains a plurality of original template palmprint images, which are collected in advance, such as collected by a web crawler or manually collected; therefore, the target template palmprint image set contains a plurality of target Template palm print image.

Preferably, as shown in Figure 2, in S2, the specific steps of obtaining the target template palmprint image set include:

S2a.1: respectively extracting the region of interest of each original template palmprint image in the original template palmprint image set to obtain a template palmprint ROI image set;

S2a.2: Using the MFRAT filtering method, each template palmprint ROI image in the template palmprint ROI image set is filtered separately to obtain a template palmprint filtered image set;

S2a.3: In each template palmprint filtering image of the template palmprint filtering image set, encode each pixel point respectively to obtain the target template palmprint image set;

As shown in Figure 3, in S2, the concrete steps that obtain described target palmprint image include:

S2b.1: Extract the region of interest of the original palmprint image to obtain a palmprint ROI image;

S2b.2: Using the MFRAT filtering method, filter the palmprint ROI image to obtain a palmprint filtered image;

S2b.3: In the palmprint filtered image, encode each pixel point separately to obtain the target palmprint image.

Because the region of interest (Regions of Interest, ROI) contains more effective palmprint features, so first extract the region of interest of each original template palmprint image in the original template palmprint image set and the region of interest of the original palmprint image, It can not only extract effective palmprint features in each palmprint image, but also facilitate subsequent feature extraction and palmprint recognition matching, and can also greatly reduce the amount of calculation; MFRAT (Modified Finite Radon Transform) filtering method is an improved finite Radon Transformation can accurately locate the palmprint features. Therefore, the filtering process of the MFRAT filtering method can effectively avoid the influence of noise on palmprint recognition. Each pixel in the palmprint filtering image of each template and the palmprint filtering image Each pixel in the palmprint is encoded separately, which is convenient for the subsequent feature extraction and the execution of the image pyramid method, so as to facilitate the recognition and matching of palmprints.

Preferably, S2a.2 specifically includes:

S2a.2.1: Perform histogram equalization processing on each template palmprint ROI image in the template palmprint ROI image set to obtain a first intermediate template palmprint image set;

S2a.2.2: Perform normalization processing on each of the first intermediate template palmprint images in the first intermediate template palmprint image set to obtain a second intermediate template palmprint image set;

S2a.2.3: In each of the second intermediate template palmprint images in the second intermediate template palmprint image set, construct the first MFRAT filter function, take any pixel point as the first center point, and establish p×p The first filter grid, and in the first filter grid, according to the first MFRAT filter function, calculate and obtain the first center point in each direction of a plurality of first response values, and according to all the first response values in each direction The first response value obtains the first pixel cumulative value of the first central point in each direction;

The first MFRAT filter function of the first center point (x, y) in the θ _k direction is:

Wherein, (x, y) is the coordinate of the first central point in the second intermediate template palmprint image, and r (x, y) is the pixel value corresponding to the first central point (x, y), θ _k (k=1 ,2,…,6) are the selected six directions, which are 0, π/6, 2π/6, 3π/6, 4π/6 and 5π/6, is a straight line equation of the first filtering grid in the direction of θ _k , is the straight line equation of the first center point (x, y) in the direction of θ _k The first response value on ;

S2a.2.4: traverse each pixel of the second intermediate template palmprint image, according to the method of step 2a.2.3, obtain the first pixel cumulative value of each pixel in each direction, and according to All the first pixel cumulative values of all pixels are obtained corresponding to a template palmprint filter image corresponding to the second intermediate template palmprint image;

S2a.2.5: Obtain the template palmprint filtered image set according to all template palmprint filtered images;

S2b.2 specifically includes:

S2b.2.1: Perform histogram equalization processing on the palmprint ROI image to obtain a first intermediate palmprint image;

S2b.2.2: Perform normalization processing on the first intermediate palmprint image to obtain a second intermediate palmprint image;

S2b.2.3: In the second intermediate palmprint image, construct a second MFRAT filter function, take any pixel point as the second center point, establish a p×p second filter grid, and In the grid, according to the MFRAT filter function, a plurality of second response values of the second center point in each direction are calculated, and according to all second response values in each direction, the second center point is obtained in each direction. The accumulated value of the second pixel in the direction;

The second MFRAT filter function of the second central point (x′, y′) in the direction of θ _k is:

Wherein, (x', y') is the coordinate of the second center point in the second middle palmprint image, and r'(x', y') is the pixel value corresponding to the second center point (x', y'), is a straight line equation of the second filtering grid in the θ _k direction, is the straight line equation of the second center point (x′,y′) in the direction of θ _k The second response value on ;

S2b.2.4: traverse each pixel of the second intermediate palmprint image, and obtain the second pixel accumulation value of each pixel in each direction according to the method of step 2b.2.3;

S2b.2.5: Obtain the palmprint filtered image according to the accumulated values of all second pixels of all pixels.

Through the filtering process of the above steps, on the one hand, the influence of noise on palmprint recognition can be effectively overcome, and on the other hand, the contrast of the palmprint features in each image can be enhanced, so as to facilitate the execution of the subsequent SIFT method and image pyramid method, even in The accurate target SIFT feature point set and the template feature point set are extracted, and the accurate offset between the target palmprint image and each target template palmprint image is conveniently calculated.

Specifically, one of the template palmprint ROI images I (x, y) of the template palmprint ROI image set in the present embodiment has a size of M × N (M and N are respectively the total number of row pixels of the template palmprint ROI image and Column total number of pixels); This template palmprint ROI image I (x, y) is carried out histogram equalization processing, obtains the corresponding first intermediate template palmprint ROI image as I _H (x, y), then the first Middle template palmprint ROI image is I _H (x, y) carries out normalization process, obtains the second middle template palmprint ROI image after normalization is I _N (x, y); , y) is the first center point, construct the first filter grid of 11×11, in the first filter grid, filter according to the first MFRAT filter function, and calculate the first center point in six directions The cumulative value of the first pixel on M _i , where i=1,2,...,6; traverse the second intermediate template palmprint ROI image as each pixel in I _N (x,y), according to the same The method calculates the one-to-one corresponding first pixel cumulative value of each pixel, and obtains the second intermediate template palmprint ROI image corresponding to I _N (x, y) in six directions according to all the first pixel cumulative values. Template palmprint filter image is I _Mi (x, y), and wherein i=1,2,...,6; According to above-mentioned same method, can obtain template palmprint ROI image set each template palmprint ROI image corresponding in six Template palmprint filtering images in three directions, and palmprint filtering images J _Mi (x′, y′) corresponding to the palmprint ROI image, where i=1, 2, . . . , 6.

Preferably, S2a.3 specifically includes:

S2a.3.1: In each template palmprint filtering image in the template palmprint filtering image set, use the direction corresponding to the maximum value of all first pixel cumulative values of each pixel as the first feature of the corresponding pixel coded value;

S2a.3.2: According to all the first feature coding values corresponding to all pixels in each template palmprint filtering image, obtain a one-to-one corresponding first feature coding value subset of each template palmprint filtering image;

S2a.3.3: Obtain the target template palmprint image set according to all first feature coding value subsets corresponding to all template palmprint filtered images;

S2b.3 specifically includes:

S2b.3.1: In the palmprint filtered image, use the direction corresponding to the maximum value of all the second pixel cumulative values of each pixel as the second feature encoding value of the corresponding pixel;

S2b.3.2: Obtain a subset of second feature code values corresponding to the palmprint filter image according to all second feature code values corresponding to all pixels in the palmprint filter image;

S2b.3.3: Obtain the target palmprint image according to the second feature code value subset.

Through the encoding method of the above steps, it is convenient to design corresponding feature matching strategies according to the target template palmprint image set and target palmprint image obtained after encoding, so as to facilitate follow-up in the target template palmprint image set and target palmprint image according to the feature matching strategy Feature extraction is carried out, and feature matching is carried out according to the feature matching strategy, which effectively improves the accuracy and precision of palmprint recognition and matching, and the effect of recognition and matching is good.

Specifically, in the present embodiment, for the second intermediate template palmprint ROI image, the template palmprint filter image corresponding to I _N (x, y) in six directions is I _Mi (x, y), and at any pixel point Above, select the maximum value among the accumulated values of the first pixels in the 6 directions, and the corresponding direction is the first feature code value of the pixel, traverse each pixel, and obtain the palmprint ROI image of the second intermediate template as I _N (x, y) corresponds to the first feature encoding value subset, and then obtains the target template palmprint image I _M (x, y) according to the first feature encoding value subset; adopt the above-mentioned same method to obtain each The target template palmprint image corresponding to the second intermediate template palmprint ROI image, and the target palmprint image J _M (x′, y′) corresponding to the second intermediate palmprint ROI image, all target template palmprint images constitute the target Set of template palm print images.

Preferably, as shown in Figure 4, in S3, the specific steps of extracting the one-to-one template feature point set of each target template palmprint image in the target template palmprint image set include:

S3a.1: Using the bilinear interpolation method, expand the palmprint image of each target template respectively to obtain the enlarged target template palmprint image set;

S3a.2: In each enlarged target template palmprint image in the expanded target template palmprint image set, use the scale-invariant feature transformation method to construct the first scale space, and detect the A set of first extremum points corresponding to each expanded target template palmprint image in the first scale space;

S3a.3: Use the Harris Comer detector to filter the first extreme point set of each expanded target template palmprint image in the first scale space to obtain a one-to-one corresponding template for each target template palmprint image feature point set;

As shown in Figure 5, in S3, the specific steps of extracting the target SIFT feature point set of the target palmprint image include:

S3b.1: Enlarging the target palmprint image by using a bilinear interpolation method to obtain the enlarged target palmprint image;

S3b.2: In the expanded target palmprint image set, use a scale-invariant feature transformation method to construct a second scale space, and detect that the expanded target palmprint image is in the second scale space according to the preset second pixel threshold The corresponding second extreme point set in the two-scale space;

S3b.3: Using a Harris Comer detector to filter all the second extreme points in the second extreme point set to obtain the target SIFT feature point set corresponding to the target palmprint image.

Expand each target template palmprint image and target palmprint image by bilinear interpolation method, so as to facilitate subsequent detection of the corresponding first extreme point set and second extreme value using the scale invariant feature change method (SIFT method) point set; the Harris Comer detection (Harris corner point detection) method is a point feature extraction method based on the signal, so the first extreme point set and the second extreme point set are filtered respectively by the Harris Comer detector, which can be filtered Get rid of some unstable feature points to get template feature point set and target SIFT feature point set with high stability; among them, the preset pixel threshold can be determined and adjusted according to the actual situation, and the Harris Comer detector is used for filtering. The operation steps are relatively mature technologies at present, and will not be repeated here.

It should be noted that, through the filtering of the Harris Comer detector, the set of template feature points corresponding to each enlarged target template palmprint image is firstly obtained, but since each expanded target template palmprint image is composed of the corresponding target template palmprint image The palmprint image is obtained by expanding, so the template feature point set corresponding to each expanded target template palmprint image is also the template feature point set corresponding to each target template palmprint image; The target SIFT feature point set corresponding to the palmprint image is also the target SIFT feature point set corresponding to the target palmprint image.

Specifically, in this embodiment, for the target template palmprint image I _M (x, y), its size is M × N, and the bilinear interpolation method is adopted, and the enlarged target template palmprint image obtained is I _MD (x, y ), its size is 2M * 2N, in like manner, each target template palmprint image and target palmprint image are enlarged respectively, obtain the expanded target template palmprint image corresponding to each target template palmprint image one by one, and The enlarged target palmprint image corresponding to the target palmprint image is J _MD (x′,y′), wherein all the enlarged target template palmprint images constitute the enlarged target template palmprint image set.

Specifically, in this embodiment, in the enlarged target template palmprint image I _MD (x, y), a scale-invariant feature transformation method is used to construct a first scale space, which includes S scales and satisfies Among them, k is the scale factor coefficient, S is set to 3, that is, a three-layer tower image is constructed, and in the DOG scale space of each layer, the pixel gray value of each pixel is detected according to the preset first pixel threshold Whether it is the first extremum point in the local space range, detect all the first extremum points; and use the Harris Comer detector to filter the unstable first extremum point, and obtain the relatively stable one corresponding to the expanded target template palmprint image The template feature point set is P _Ij1 , where j ₁ =1,2,...,a ₁ (a ₁ is the total number of feature points in the template feature point set corresponding to the expanded target template palmprint image); similarly, each enlarged The relatively stable template feature point set corresponding to the target template palmprint image one-to-one (that is, the relatively stable template feature point set corresponding to each target template palmprint image), and the relatively stable template feature point set corresponding to the expanded target palmprint image. Target SIFT feature point set P _Jj (that is, the relatively stable target SIFT feature point set corresponding to the target palmprint image), where j=1, 2,..., b (b is the target SIFT feature point corresponding to the enlarged target palmprint image The total number of feature points in the set).

Specifically, in this embodiment, after obtaining the above-mentioned relatively stable template feature point set and relatively stable target SIFT feature point set, the second filtering operation can be performed respectively, and for each feature point, the surrounding 4 pixels are filtered out The other feature points of the model are obtained respectively to obtain the most stable template feature point set and the most stable target SIFT feature point set.

Preferably, as shown in Figure 6, the specific steps of S4 include:

S4.1: Based on the image pyramid method, according to each target template palmprint image and the corresponding template feature point set, obtain the one-to-one corresponding template corner point coordinate set of each target template palmprint image, according to the target palmprint image With the target SIFT feature point set, obtain the target corner point coordinate set corresponding to the target palmprint image;

S4.2: Using the BLPOC method, according to each template corner point coordinate set and the target corner point coordinate set, respectively calculate the offset between the target palmprint image and each target template palmprint image.

Based on the image pyramid method, the image pyramid is constructed, and in the image pyramid, the set of template corner coordinates corresponding to each target template palmprint image and the set of target corner coordinates corresponding to the target palmprint image are respectively obtained. The palmprint deformation is corrected in a local area, which facilitates the subsequent calculation of the offset between the target palmprint image and each target template palmprint image, especially for the low-resolution palmprint recognition process, which effectively overcomes the problem of palmprint deformation. The influence of deformation on palmprint recognition can further improve the recognition accuracy of palmprint; the BLPOC method (Band-limited Phase-only Correlation, band-limited phase correlation method) can effectively extract the phase features of the knuckle image, through the cross-power spectrum of the knuckle image Therefore, the offset between each template corner point coordinate set and the target corner point coordinate set can be calculated more accurately through the BLPOC method, that is, a more accurate target palmprint image can be obtained The offset between the palmprint image and each target template, so as to facilitate subsequent recognition and matching of the palmprint according to the offset.

Specifically, in this embodiment, based on the image pyramid method and the principle of image pyramid calculation, a two-layer image pyramid is constructed. For the palmprint image of the target template, the palmprint image of the target template is first sampled to obtain a target template with a scale of 1/2 For the palmprint image, use the palmprint ROI (Region of Interest) segmentation method (such as the Otsu threshold segmentation algorithm) to segment the 1/2-scale target template palmprint image, and determine the sampling points for each segmented block according to the degree of variation , this is because in general, it is assumed that the sampling points in the palmprint image are evenly distributed, but the actual distribution of the palmprint lines is not fully considered in this assumption, and the palmprint texture is a relatively complex random structure, evenly distributed sampling points can neither accurately collect palmprint information, but also lead to the formation of redundant collection points; therefore, the sampling points of each segment are selected according to the degree of variation, and the number of sampling points is determined according to the degree of variation. The specific position can improve the accuracy of palmprint recognition; then determine the corner coordinates in each segment, and obtain the corresponding corner coordinates as T _It1 , t ₁ =1,2,...,c ₁ , where c ₁ represents the corner Quantity, with the corner point T _I1 as a reference point, according to the double-layer image pyramid algorithm, the offset between the feature points corresponding to the target palmprint image corresponding to the palmprint to be matched is O ₁ ; each corner point is traversed , calculate its corresponding offset, denoted as O _f , t ₁ =1,2,...,c ₁ ; similarly, calculate the corresponding target palmprint image corresponding to each target template palmprint image and the palmprint to be matched The offset between the feature points, that is, the offset between each original template palmprint image and the original palmprint image; among them, constructing an image pyramid, image segmentation, selecting sampling points according to the degree of variation, and obtaining the template The specific operation steps of the corner point coordinate set and the target corner point coordinate set acquisition are all in the prior art, and the specific operation steps of the BLPOC method are also in the prior art, and the specific details will not be repeated here.

Preferably, the specific implementation of S5 is:

Taking the target template palmprint image corresponding to the minimum value of all offsets as the recognition result of the palmprint to be matched and outputting it.

When the offset is the smallest, it means that the target template palmprint image corresponding to the minimum value of the offset is the closest to the target palmprint image corresponding to the palmprint to be matched, that is, the palmprint image of the target template is the palmprint image corresponding to the palmprint to be matched. Output the palmprint image that matches the target palmprint image the most, and then output the palmprint image of the target template to obtain the recognition result with the highest recognition accuracy, and the recognition effect is good.

Specifically, since in this embodiment, each target template image has a plurality of offsets corresponding to the target palmprint image corresponding to the palmprint to be matched, any target template palmprint image and the palmprint to be matched can be All the offsets corresponding to the target palmprint image corresponding to the pattern are averaged (such as averaging or weighted average), and the average value obtained is used as the mean shift between the corresponding target template palmprint image and the target palmprint image amount, and then match according to the mean offset, specifically, the target template palmprint image corresponding to the smallest mean offset is used as the recognition result of the target palmprint image (that is, the original palmprint of the palmprint to be matched The recognition result of the image) and output; wherein, the specific operation method of averaging and outputting the recognition result is an existing mature technology, and will not be repeated here.

Embodiment two, as shown in Figure 7, a kind of palmprint recognition system based on image pyramid, comprises acquisition module, preprocessing module, extraction module, calculation module and identification module:

The acquisition module is used to acquire the original palmprint image of the palmprint to be matched;

The preprocessing module is used to preprocess each original template palmprint image in the preset original template palmprint image set to obtain the target template palmprint image set, and is also used to preprocess the original palmprint image. Process to obtain the target palmprint image;

The extraction module is used to extract the target SIFT feature point set of the target palmprint image and the one-to-one corresponding template feature points of each target template palmprint image in the target template palmprint image set by adopting the scale-invariant feature transformation method set;

The calculation module is used to calculate the offset information between the target SIFT feature point set and the one-to-one corresponding template feature point set of each target template palmprint image based on the image pyramid method;

The identification module is configured to identify and match the palmprint to be matched according to all offset information in the palmprint image set of the target template, obtain and output the identification result.

Obtain the original palmprint image of the palmprint to be matched through the acquisition module, and then preprocess each original template palmprint image through the preprocessing module to obtain one-to-one corresponding target template palmprint images. These target template palmprint images constitute The target template palmprint image set is obtained, and the original palmprint image is also preprocessed to obtain the target palmprint image. Through preprocessing, the palmprint line features affected by noise can be filtered out, and the subsequent scale-invariant feature transformation is improved. The noise resistance of the method effectively overcomes the influence of noise on palmprint recognition, and also strengthens the contrast between each target template palmprint image and the texture feature of the target palmprint image, which is beneficial to subsequent palmprint recognition and matching; The target SIFT feature point set extracted by the extraction module through the SIFT method corresponds to the template feature point set of each target template palmprint image, which is further beneficial to the subsequent palmprint recognition and matching, and overcomes the impact of palmprint translation on the palmprint. The impact of recognition has significantly improved the recognition accuracy of palmprint; after extracting the template feature point set corresponding to the target SIFT feature point set and each target template palmprint image, the image pyramid method can be used through the calculation module, and the palmprint The deformation is effectively corrected in the local area, especially for the low-resolution palmprint recognition process, which effectively overcomes the influence of palmprint deformation on palmprint recognition. The method is flexible and simple, which is conducive to the subsequent calculation of the target palmprint image and each target The offset between the template palmprint images further improves the recognition accuracy of the palmprint; finally, the identification module can judge which target template palmprint image in the target palmprint image and the target template palmprint image set is the most Matching, thereby identifying the palmprint to be matched, the recognition and matching method is relatively simple and effective, and the recognition accuracy of the palmprint is obviously improved.

Preferably, as shown in FIG. 8, the preprocessing module includes a region of interest extraction unit, a filtering unit and a coding unit;

The region of interest extraction unit is used to extract the region of interest of each original template palmprint image in the original template palmprint image set respectively to obtain the template palmprint ROI image set; it is also used to extract the original palmprint image The region of interest of the palmprint ROI image is obtained;

Described filtering unit is used for adopting MFRAT filter method, each template palmprint ROI image of described template palmprint ROI image set is carried out filter processing respectively, obtains template palmprint filter image set; Also for adopting MFRAT filter method, for The palmprint ROI image is filtered to obtain a palmprint filtered image;

The encoding unit is used to encode each pixel in each template palmprint filter image of the template palmprint filter image set to obtain the target template palmprint image set; In the palmprint filtered image, each pixel is coded separately to obtain the target palmprint image.

First extract the region of interest of each original template palmprint image and the region of interest of the original palmprint image in the original template palmprint image set by the interest extraction unit, both can extract the effective palmprint feature in each palmprint image, facilitate Subsequent feature extraction and palmprint recognition matching can also greatly reduce the amount of computation; through the filtering unit, the influence of noise on palmprint recognition can be effectively avoided, and each pixel in the palmprint filtering image of each template is filtered through the encoding unit Each pixel in the palmprint filtered image is coded separately, which is convenient for the subsequent feature extraction and the execution of the image pyramid method, so as to facilitate the recognition and matching of the palmprint.

Embodiment three, based on embodiment one and embodiment two, present embodiment also discloses a kind of palmprint recognition device based on image pyramid, comprises processor, memory and is stored in described memory and can run on described processor The computer program on the computer program, when the computer program runs, realizes the specific steps of S1 to S5 as shown in FIG. 1 .

The palmprint recognition of the present invention is realized by the computer program stored in the memory and running on the processor. Based on the scale-invariant feature transformation method and the image pyramid, the palmprint deformation is effectively corrected, and the palmprint deformation and palmprint deformation are overcome. The impact of pattern translation and noise on palmprint recognition has significantly improved the recognition accuracy of palmprint, and the recognition effect is good.

This embodiment also provides a computer storage medium, where at least one instruction is stored on the computer storage medium, and the specific steps of S1 to S5 are implemented when the instruction is executed.

By executing the computer storage medium containing at least one instruction, the palmprint recognition of the present invention is realized, based on the scale-invariant feature transformation method and the image pyramid, the palmprint deformation is effectively corrected, and the palmprint deformation, palmprint translation and noise are overcome. The impact of palmprint recognition has significantly improved the recognition accuracy of palmprint, and the recognition effect is good.

For the details of S1 to S5 in this embodiment, please refer to the content of Embodiment 1 and FIG. 1 to FIG. 6 for details, and details are not repeated here.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection of the present invention. within range.

Claims (10)

1. A palm print identification method based on an image pyramid is characterized by comprising the following steps:

step 1: acquiring an original palm print image of a palm print to be matched;

step 2: respectively preprocessing each original template palm print image in a preset original template palm print image set to obtain a target template palm print image set, and preprocessing the original palm print images to obtain a target palm print image;

and step 3: respectively extracting a target SIFT feature point set of the target palm print image and a template feature point set corresponding to each target template palm print image in the target template palm print image set one by adopting a scale invariant feature transformation method;

and 4, step 4: calculating to obtain the offset between each target template palm print image and each target template palm print image according to each target template palm print image, the template feature point set corresponding to each target template palm print image one by one, the target palm print image and the target SIFT feature point set corresponding to the target palm print image based on an image pyramid method;

and 5: and in the target template palm print image set, identifying and matching the palm prints to be matched according to all the offsets to obtain and output an identification result.

2. The image pyramid-based palm print recognition method of claim 1, wherein in the step 2, the specific step of obtaining the target template palm print image set comprises:

step 2 a.1: respectively extracting the interested region of each original template palm print image in the original template palm print image set to obtain a template palm print ROI image set;

step 2 a.2: respectively filtering each template palm print ROI image in the template palm print ROI image set by adopting an MFRAT filtering method to obtain a template palm print filtering image set;

step 2 a.3: in each template palm print filtering image of the template palm print filtering image set, respectively encoding each pixel point to obtain the target template palm print image set;

in the step 2, the specific step of obtaining the target palm print image includes:

step 2 b.1: extracting an interested area of the original palm print image to obtain a palm print ROI image;

step 2 b.2: filtering the palm print ROI image by adopting an MFRAT filtering method to obtain a palm print filtering image;

step 2 b.3: and respectively coding each pixel point in the palm print filtering image to obtain the target palm print image.

3. The image pyramid-based palm print recognition method of claim 2, wherein the step 2a.2 specifically comprises:

step 2 a.2.1: respectively carrying out histogram equalization processing on each template palm print ROI image in the template palm print ROI image set to obtain a first intermediate template palm print image set;

step 2 a.2.2: respectively carrying out normalization processing on each first intermediate template palm print image in the first intermediate template palm print image set to obtain a second intermediate template palm print image set;

step 2 a.2.3: in each second intermediate template palm print image of the second intermediate template palm print image set, constructing a first MFRAT filter function, establishing a p × p first filter grid by taking any pixel point as a first central point, calculating to obtain a plurality of first response values of the first central point in each direction according to the first MFRAT filter function in the first filter grid, and obtaining a first pixel accumulated value of the first central point in each direction according to all the first response values in each direction;

the first center point (x, y) is at theta_kThe first MFRAT filter function in the direction is:

wherein (x, y) isCoordinates of a first central point in the two middle template palm print images, r (x, y) is a pixel value corresponding to the first central point (x, y), and theta_k(k-1, 2, …,6) is selected from six directions of 0, pi/6, 2 pi/6, 3 pi/6, 4 pi/6 and 5 pi/6,for the first filter grid at theta_kAn equation of a straight line in the direction,is the first center point (x, y) at theta_kEquation of straight line in directionA first response value of;

step 2 a.2.4: traversing each pixel point of each second intermediate template palm print image, obtaining a first pixel accumulated value of each pixel point in each direction according to the method of the step 2a.2.3, and obtaining a template palm print filtering image corresponding to one corresponding second intermediate template palm print image according to all the first pixel accumulated values of all the pixel points;

step 2 a.2.5: obtaining a template palmprint filtering image set according to all template palmprint filtering images;

the step 2b.2 specifically comprises:

step 2 b.2.1: performing histogram equalization processing on the palm print ROI image to obtain a first middle palm print image;

step 2 b.2.2: normalizing the first intermediate palm print image to obtain a second intermediate palm print image;

step 2 b.2.3: in the second intermediate palm print image, constructing a second MFRAT filtering function, establishing a p × p second filtering grid by taking any pixel point as a second central point, calculating to obtain a plurality of second response values of the second central point in each direction according to the second MFRAT filtering function in the second filtering grid, and obtaining a second pixel accumulated value of the second central point in each direction according to all the second response values in each direction;

the second center point (x ', y') is at theta_kThe second MFRAT filter function in the direction is:

wherein, (x ', y ') is the coordinate of the second central point in the second intermediate palm print image, r ' (x ', y ') is the pixel value corresponding to the second central point (x ', y '),for the second filter grid at θ_kAn equation of a straight line in the direction,is the second center point (x ', y') is at theta_kEquation of straight line in directionA second response value of;

step 2 b.2.4: traversing each pixel point of the second middle palm print image, and obtaining a second pixel accumulated value of each pixel point in each direction according to the method of the step 2 b.2.3;

step 2 b.2.5: and obtaining the palm print filtering image according to all the second pixel accumulated values of all the pixel points.

4. The image pyramid-based palm print recognition method of claim 3, wherein the step 2a.3 specifically comprises:

step 2 a.3.1: in each template palm print filtering image in the template palm print filtering image set, taking the direction corresponding to the maximum value in all the first pixel accumulated values of each pixel point as the first characteristic coding value of the corresponding pixel point;

step 2 a.3.2: obtaining a first feature coding value subset corresponding to each template palm print filtering image one by one according to all first feature coding values corresponding to all pixel points in each template palm print filtering image;

step 2 a.3.3: obtaining the target template palm print image set according to all the first feature coding value subsets corresponding to all the template palm print filtering images;

the step 2b.3 specifically comprises:

step 2 b.3.1: in the palm print filtering image, taking the direction corresponding to the maximum value in all the second pixel accumulated values of each pixel point as a second characteristic coding value of the corresponding pixel point;

step 2 b.3.2: obtaining a second feature code value subset corresponding to the palm print filtering image according to all second feature code values corresponding to all pixel points in the palm print filtering image;

step 2 b.3.3: and obtaining the target palm print image according to the second characteristic coding value subset.

5. The image pyramid-based palm print recognition method of claim 4, wherein in the step 3, the specific step of extracting the template feature point set corresponding to each target template palm print image in the target template palm print image set one to one comprises:

step 3 a.1: respectively expanding each target template palm print image by adopting a bilinear interpolation method to obtain an expanded target template palm print image set;

step 3 a.2: in each enlarged target template palm print image of the enlarged target template palm print image set, adopting a scale invariant feature transformation method to construct a first scale space, and detecting a first extreme point set corresponding to each enlarged target template palm print image in the first scale space one by one according to a preset first pixel threshold;

step 3 a.3: filtering a first extreme point set of each enlarged target template palm print image in the first scale space by adopting a Harris Commer detector to obtain a template feature point set corresponding to each target template palm print image one to one;

in the step 3, the specific step of extracting the target SIFT feature point set of the target palm print image includes:

step 3 b.1: expanding the target palm print image by adopting a bilinear interpolation method to obtain an expanded target palm print image;

step 3 b.2: constructing a second scale space in the enlarged target palm print image set by adopting a scale-invariant feature transformation method, and detecting a corresponding second extreme point set of the enlarged target palm print image in the second scale space according to a preset second pixel threshold;

step 3 b.3: and filtering all second extreme points in the second extreme point set by adopting a Harris comber detector to obtain the target SIFT feature point set corresponding to the target palm print image.

6. The image pyramid-based palm print recognition method of claim 5, wherein the specific steps of the step 4 include:

step 4.1: acquiring a template corner point coordinate set corresponding to each target template palm print image one by one according to each target template palm print image and a corresponding template feature point set based on an image pyramid method, and acquiring a target corner point coordinate set corresponding to the target palm print image according to the target palm print image and the target SIFT feature point set;

step 4.2: and respectively calculating the offset between the target palm print image and each target template palm print image according to each template corner point coordinate set and the target corner point coordinate set by adopting a BLPOC (global binary offset plus one minus one plus.

7. The image pyramid-based palm print recognition method of claim 6, wherein the step 5 is implemented by:

and taking the target template palm print image corresponding to the minimum value in all the offsets as the identification result of the palm print to be matched and outputting the identification result.

8. The utility model provides a palm print identification system based on image pyramid which characterized in that, includes acquisition module, preprocessing module, draws module, calculation module and identification module:

the acquisition module is used for acquiring an original palm print image of a palm print to be matched;

the preprocessing module is used for respectively preprocessing each original template palm print image in a preset original template palm print image set to obtain a target template palm print image set, and is also used for preprocessing the original palm print images to obtain a target palm print image;

the extraction module is used for respectively extracting a target SIFT feature point set of the target palm print image and a template feature point set corresponding to each target template palm print image in the target template palm print image set by adopting a scale invariant feature transformation method;

the calculation module is used for calculating to obtain offset information between the target SIFT feature point set and the template feature point set corresponding to each target template palm print image one by one based on an image pyramid method;

and the identification module is used for identifying and matching the palm print to be matched according to all the offset information in the target template palm print image set to obtain an identification result and outputting the identification result.

9. An image pyramid-based palm print recognition apparatus comprising a processor, a memory and a computer program stored in the memory and executable on the processor, the computer program when executed implementing the method steps according to any one of claims 1 to 7.

10. A computer storage medium, the computer storage medium comprising: at least one instruction which, when executed, implements the method steps of any one of claims 1 to 7.