Fingerprint image matching and recognition method, device, electronic device and storage medium

Abstract

The embodiment of the present invention discloses a fingerprint image matching and recognition method, device, electronic device and storage medium, which relates to the field of machine vision and biometrics, wherein the method comprises: obtaining a fingerprint image to be matched, performing position preprocessing and candidate point verification to obtain feature points, using a circular domain angle voting method to calculate the non-polarity direction of the feature points, obtaining hash features of the feature points by downsampling mirror difference, hash mapping and symbol quantization, selecting candidate pairing points by calculating the Hamming similarity between the feature points and sorting them, calculating the matching transformation relationship between the candidate pairing points by a random clustering strategy and verifying the number of pairing points, and successfully matching if the preset conditions are met. The present invention extracts the origin symmetric hash features in the fingerprint image and designs an efficient feature matching algorithm, so that the feature description is more comprehensive and accurate, and solves the problem that the prior art is easily affected by noise and acquisition conditions, has low recognition efficiency and low matching accuracy.

Description

Technical Field

[0001] The present invention relates to the field of machine vision and biometrics, and in particular to a fingerprint image matching and recognition method, device, electronic equipment and storage medium. Background Art

[0002] In the field of machine vision and biometrics, fingerprint recognition, as an efficient and convenient means of identity authentication, has been widely used in multiple scenarios such as mobile phone unlocking, access control systems, and payment verification. With the continuous development of semiconductor fingerprint acquisition sensor technology, the acquisition and processing of small-area fingerprint images has become particularly important, and small-area fingerprint recognition technology is widely used in scenarios with strict space requirements or high portability requirements, such as smartphones, wearable devices, smart cards, etc. These devices usually need to complete user identity authentication quickly and accurately.

[0003] However, when it comes to the recognition of small-area fingerprint images, traditional fingerprint recognition algorithms often have problems with low recognition efficiency and low matching accuracy due to the limited amount of information they contain and their susceptibility to noise and acquisition conditions. This is especially evident in the presence of noise interference and poor image quality.

[0004] In addition, small-area fingerprint images contain less effective information, which increases the difficulty of recognition. Noise may be introduced during the sensor acquisition process, affecting the image quality. Traditional algorithms often need to perform complex calculations when processing small-area fingerprint images, resulting in a decrease in recognition speed. When there are fewer feature points or the image quality is poor, the matching accuracy is easily affected.

[0005] Therefore, there is an urgent need for a fingerprint image matching and recognition method that can effectively suppress noise interference, reduce the amount of calculation, and improve recognition efficiency and matching accuracy based on the characteristics of small-area fingerprint images. Summary of the invention

[0006] The embodiments of the present invention provide a fingerprint image matching and recognition method to solve the problem that the prior art is easily affected by noise and acquisition conditions, has low recognition efficiency and low matching accuracy. The technical solution is as follows:

[0007] According to one aspect of the present invention, a fingerprint image matching and recognition method is provided, the method comprising: obtaining a fingerprint image to be matched, performing position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, and obtaining feature points according to the relative positions of the candidate points; using a circular domain angle voting method to calculate the non-polarity direction of the feature points, and obtaining hash features of the feature points through downsampling mirror difference, hash mapping and symbol quantization; calculating the Hamming similarity between each feature point in the feature points of the fingerprint image to be matched and sorting them, and selecting the feature points that meet the set conditions as candidate pairing points; calculating the matching transformation relationship between the candidate pairing points through a random clustering strategy and verifying the number of pairing points, if the number of pairing points meets the preset conditions, the fingerprint image to be matched is successfully matched.

[0008] In one of the embodiments, the fingerprint image to be matched is subjected to position preprocessing and candidate point verification to obtain candidate points, and characteristic points are obtained according to the relative positions of the candidate points by the following steps: noise suppression is performed on the fingerprint image to be matched by Gaussian filtering, and the Sobel gradient and gradient amplitude of the fingerprint image to be matched are calculated; foreground and background segmentation is performed on the fingerprint image to be matched by the gradient amplitude and a preset threshold to obtain a fingerprint foreground area, and the verification vector of each pixel point in the fingerprint foreground area is calculated; candidate points are obtained according to the verification vector of the pixel point, and the candidate points are screened by neighborhood difference and 8-neighborhood adjacent conditions to obtain characteristic points.

[0009] In one of the embodiments, the foreground and background segmentation of the fingerprint image to be matched is performed using the gradient amplitude and a preset threshold to obtain the fingerprint foreground area through the following steps: the gradient amplitude is subjected to a mean filter with a radius of 7 to obtain the amplitude mean of each pixel point of the fingerprint image to be matched; if the amplitude mean of the pixel point is not less than the set threshold, it is marked as the fingerprint foreground area; the set threshold is obtained by the imaging distribution of the acquisition sensor of the fingerprint image to be matched.

[0010] In one of the embodiments, candidate points are obtained according to the check vector of the pixel point, and the candidate points are screened by neighborhood difference and 8-neighborhood neighbor conditions to obtain feature points, which is achieved by the following steps: if the number of components greater than or equal to 0 in the check vector of the pixel point is greater than or equal to 3, the pixel point is marked as a maximum candidate point; if the number of components less than or equal to 0 in the check vector of the pixel point is greater than or equal to 3, the pixel point is marked as a minimum candidate point; if two of the minimum candidate points meet the 8-neighborhood neighbor conditions, the neighborhood difference of the two minimum candidate points is calculated, and the minimum candidate point with the smallest neighborhood difference is used as a feature point; if two of the maximum candidate points meet the 8-neighborhood neighbor conditions, the neighborhood difference of the two maximum candidate points is calculated, and the maximum candidate point with the largest neighborhood difference is used as a feature point.

[0011] In one of the embodiments, the circular domain angle voting method is used to calculate the non-polarity direction of the feature point through the following steps: the circular domain angle voting method is used in combination with the direction information of the surrounding pixel points to calculate the non-polarity direction of each feature point; the non-polarity direction of each feature point is subjected to weighted histogram statistics and smoothing processing to optimize the non-polarity direction.

[0012] In one of the embodiments, obtaining the hash feature of the feature point by downsampling mirror difference, hash mapping and symbolic quantization is achieved by the following steps: constructing a square feature domain with the feature point as the center, and calculating the central symmetric feature vector of the square feature domain by using the downsampling mirror difference method; using the Hadamard matrix to hash map the feature vector, and obtaining the hash feature of the feature point by symbolic quantization.

[0013] In one of the embodiments, calculating the matching transformation relationship between the candidate pairing points through a random clustering strategy and verifying the number of pairing points is achieved by the following steps: calculating the transformation parameters of the similarity transformation relationship between the candidate pairing points through a random clustering strategy and setting the matching point pair relationship; the transformation parameters include a similarity ratio, a rotation angle, and a translation parameter; performing a consistency matching transformation according to the transformation parameters, verifying the matching of the candidate pairing points, and obtaining the number of pairing points.

[0014] According to one aspect of the present invention, a fingerprint image matching and recognition device comprises: a feature point acquisition module, used to acquire a fingerprint image to be matched, perform position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, and obtain feature points according to the relative positions of the candidate points; a feature extraction module, used to calculate the non-polarity direction of the feature points by using a circular domain angle voting method, and obtain hash features of the feature points by downsampling mirror difference, hash mapping and symbol quantization; a similarity calculation module, used to select the feature points that meet the set conditions as candidate pairing points by calculating the Hamming similarity between each feature point in the feature points of the fingerprint image to be matched and sorting them; and an image matching module, used to calculate the matching transformation relationship between the candidate pairing points by a random clustering strategy and verify the number of pairing points. If the number of pairing points meets the preset conditions, the fingerprint image to be matched is successfully matched.

[0015] According to one aspect of the present invention, an electronic device includes at least one processor and at least one memory, wherein the memory stores computer-readable instructions; the computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the fingerprint image matching and recognition method as described above.

[0016] According to one aspect of the present invention, a storage medium stores computer-readable instructions, and the computer-readable instructions are executed by one or more processors to implement the fingerprint image matching and recognition method as described above.

[0017] The beneficial effects brought by the technical solution provided by the present invention are:

[0018] In the above technical scheme, the present invention first obtains the fingerprint image to be matched, performs position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, obtains feature points according to the relative positions of the candidate points, uses the circular domain angle voting method to calculate the non-polarity direction of the feature points, obtains the hash features of the feature points by downsampling mirror difference, hash mapping and symbol quantization, calculates the Hamming similarity between each feature point in the feature points of the fingerprint image to be matched and sorts them, selects the feature points that meet the set conditions as candidate pairing points, calculates the matching transformation relationship between the candidate pairing points by random clustering strategy and verifies the number of pairing points, and if the number of pairing points meets the preset conditions, the fingerprint image to be matched is successfully matched, and by extracting the origin symmetric hash features in the fingerprint image and designing an efficient feature matching algorithm, not only the local details of the fingerprint image are considered, but also the global symmetry information is integrated, so that the feature description is more comprehensive and accurate, and the matching accuracy and robustness are improved. Rapid and accurate recognition of small-area fingerprint images is achieved, which not only reduces the amount of calculation, but also improves the recognition speed and matching accuracy, thereby effectively solving the problem that the prior art is easily affected by noise and acquisition conditions, has low recognition efficiency and low matching accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in describing the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative work.

[0020] Figure 1 is a flow chart of a fingerprint image matching and recognition method according to an exemplary embodiment;

[0021] Figure 2 It is a schematic diagram of the process of fingerprint image registration in an application scenario;

[0022] Figure 3 yes Figure 2 Schematic diagram of the fingerprint image matching verification process in the corresponding application scenario;

[0023] Figure 4 yes Figure 2 Schematic diagram of the feature extraction process in the corresponding application scenario;

[0024] Figure 5 yes Figure 2 Schematic diagram of the feature matching process in the corresponding application scenario;

[0025] Figure 6 is a block diagram of a fingerprint image matching and recognition device according to an exemplary embodiment;

[0026] Figure 7 is a hardware structure diagram of an electronic device according to an exemplary embodiment;

[0027] Figure 8 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0028] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0029] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0030] The present invention provides a fingerprint image matching and recognition method, which extracts the origin symmetric hash features in the fingerprint image and designs an efficient feature matching algorithm, so that the feature description is more comprehensive and accurate, and solves the problem that the prior art is easily affected by noise and acquisition conditions, has low recognition efficiency and low matching accuracy. The fingerprint image matching and recognition method is suitable for a fingerprint image matching and recognition device, which can be an electronic device. The fingerprint image matching and recognition method in the embodiment of the present invention can be applied to a variety of scenarios, such as fingerprint image matching and recognition.

[0031] See also Figure 1 , an embodiment of the present invention provides a fingerprint image matching and recognition method, which is applicable to electronic devices.

[0032] In the following method embodiments, for ease of description, the execution subject of each step of the method is taken as an electronic device as an example for illustration, but this does not constitute a specific limitation.

[0033] like Figure 1 As shown, the method may include the following steps:

[0034] Step 110, obtaining a fingerprint image to be matched, performing position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, and obtaining feature points according to the relative positions of the candidate points.

[0035] In a possible implementation, Gaussian filtering is used to suppress noise in the fingerprint image to be matched, the Sobel gradient and gradient amplitude of the fingerprint image to be matched are calculated, the foreground and background of the fingerprint image to be matched are segmented by the gradient amplitude and a preset threshold to obtain the fingerprint foreground area, the check vector of each pixel in the fingerprint foreground area is calculated, the candidate points are obtained according to the check vector of the pixel point, and the candidate points are screened by neighborhood difference and 8-neighborhood adjacent conditions to obtain feature points.

[0036] In a possible implementation, the gradient amplitude is filtered with a radius of 7 to obtain the amplitude mean of each pixel of the fingerprint image to be matched. If the amplitude mean of the pixel is not less than a set threshold, it is marked as the fingerprint foreground area.

[0037] The set threshold is obtained through the imaging distribution of the acquisition sensor of the fingerprint image to be matched.

[0038] In one possible implementation, if the number of components greater than or equal to 0 in the check vector of a pixel point is greater than or equal to 3, the pixel point is marked as a maximum candidate point; if the number of components less than or equal to 0 in the check vector of a pixel point is greater than or equal to 3, the pixel point is marked as a minimum candidate point; if two minimum candidate points satisfy the 8-neighborhood adjacency condition, the neighborhood difference of the two minimum candidate points is calculated, and the minimum candidate point with the smallest neighborhood difference is taken as the feature point; if two maximum candidate points satisfy the 8-neighborhood adjacency condition, the neighborhood difference of the two maximum candidate points is calculated, and the maximum candidate point with the largest neighborhood difference is taken as the feature point.

[0039] Step 130, using the circular domain angle voting method to calculate the non-polarized direction of the feature point, and obtaining the hash feature of the feature point through downsampling mirror difference, hash mapping and symbol quantization.

[0040] In a possible implementation, the circular angle voting method is combined with the direction information of the surrounding pixels to calculate the non-polarity direction of each feature point, and the non-polarity direction of each feature point is optimized by weighted histogram statistics and smoothing.

[0041] In one possible implementation, a square feature domain is constructed with the feature point as the center, the central symmetric feature vector of the square feature domain is calculated using the down-sampling mirror difference method, the feature vector is hash mapped using the Hadamard matrix, and the hash feature of the feature point is obtained through symbolic quantization.

[0042] Step 150 , by calculating the Hamming similarity between each feature point in the fingerprint image to be matched and sorting them, feature points that meet the set conditions are selected as candidate pairing points.

[0043] Step 170, the matching transformation relationship between the candidate pairing points is calculated by random clustering strategy and the number of pairing points is verified. If the number of pairing points meets the preset conditions, the fingerprint image to be matched is matched successfully.

[0044] In a possible implementation, a random clustering strategy is used and matching point pair relationships are set to calculate transformation parameters of similarity transformation relationships between candidate pairing points, and consistency matching transformation is performed according to the transformation parameters to verify the matching of candidate pairing points and obtain the number of pairing points.

[0045] The transformation parameters include similarity ratio, rotation angle and translation parameter.

[0046] Through the above process, the present invention first obtains the fingerprint image to be matched, performs position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, obtains feature points according to the relative positions of the candidate points, uses the circular domain angle voting method to calculate the non-polarity direction of the feature points, obtains the hash features of the feature points by downsampling mirror difference, hash mapping and symbol quantization, calculates the Hamming similarity between each feature point in the feature points of the fingerprint image to be matched and sorts them, selects the feature points that meet the set conditions as candidate pairing points, calculates the matching transformation relationship between the candidate pairing points by random clustering strategy and verifies the number of pairing points, and if the number of pairing points meets the preset conditions, the fingerprint image to be matched is successfully matched, and by extracting the origin symmetric hash features in the fingerprint image and designing an efficient feature matching algorithm, not only the local details of the fingerprint image are considered, but also the global symmetry information is integrated, so that the feature description is more comprehensive and accurate, and the accuracy and robustness of the matching are improved. Rapid and accurate recognition of small-area fingerprint images is achieved, not only the amount of calculation is reduced, but also the recognition speed and matching accuracy are improved, so that the problem that the prior art is easily affected by noise and acquisition conditions, the recognition efficiency is low, and the matching accuracy is not high can be effectively solved.

[0047] In one application scenario, Figure 2 The schematic diagram of fingerprint image registration process is shown. Figure 3 The schematic diagram of the feature extraction process is shown. Figure 4 The schematic diagram of the fingerprint image matching verification process is shown. Figure 5A schematic diagram of the feature matching process is shown.

[0048] like Figure 2 As shown in the figure, firstly, a small area fingerprint image of registered user A is input, and then the position, direction, and hash features of the feature point set are extracted from the fingerprint image as the matching feature output of registered user A. Figure 2 The following figure shows an example of an input image and a visualization of the locations of the extracted feature point sets.

[0049] like Figure 3 As shown in the figure, a small fingerprint image of unknown user B and the matching features of registered user A are input, and after feature matching, the result of whether the verification is passed is output. Figure 3 A visualization example of feature point set matching is shown in Figure .

[0050] Specifically, Figure 4 and Figure 5 They were shown Figure 2 The process of feature extraction and Figure 3 The process of feature matching in Figure 4 As shown in the figure, firstly, a small-area fingerprint image is obtained, and then the fingerprint image is noise-filtered and its gradient amplitude is calculated. The foreground area is segmented according to the gradient amplitude, and the check vector of each pixel position in the foreground area is calculated. According to the check vector, each pixel point is divided into three categories: maximum candidate point, minimum candidate point and non-candidate point. All candidate points are traversed, and the adjacent candidate points are removed according to the size of the neighborhood difference of the candidate point. The remaining candidate points are marked as feature points, and then the circular domain weighted voting method is used to calculate the non-polar direction of each feature point. The square feature domain of each feature point is calculated according to the non-polar direction. The mirror difference method is used in the square feature domain to calculate the centrally symmetric feature vector, and the feature vector is Hadamard hash mapped and quantized to obtain the hash feature. Finally, the hash feature is further optimized, and the feature point set packaged after the position, direction and hash feature of all feature points are compressed is output as the matching feature.

[0051] like Figure 5 As shown, according to the above process, the matching features of small-area fingerprint images A and B are obtained respectively, and then the hash feature Hamming distance between the two sets of feature point sets is calculated to obtain the three candidate points in the feature points of fingerprint image A that are most likely to match fingerprint image B. A random clustering strategy is used to calculate the true matching relationship between the two sets of feature point sets. During the voting process, preset matching parameters are set to accelerate the weighted voting calculation. The weighted voting process is iterated until the preset conditions are met. The consistent matching transformation relationship is used to verify the final matching pairs, and the number of point pairs on the final match is obtained. It is determined whether the number exceeds the preset threshold. If so, it indicates that the matching verification is passed, otherwise it indicates that the matching verification fails.

[0052] Through the above process, the embodiment of the present invention extracts the origin symmetric hash features in the fingerprint image and designs an efficient feature matching algorithm, which not only takes into account the local details of the fingerprint image, but also incorporates the global symmetry information, making the feature description more comprehensive and accurate, and improving the accuracy and robustness of the matching. It realizes the rapid and accurate recognition of small-area fingerprint images, which not only reduces the amount of calculation, but also improves the recognition speed and matching accuracy, thereby effectively solving the problem that the prior art is easily affected by noise and acquisition conditions, and has low recognition efficiency and low matching accuracy.

[0053] In another application scenario, it is assumed that the small-area fingerprint image to be matched obtained by the semiconductor fingerprint acquisition sensor is , whose size is At present, almost all the images collected by semiconductor fingerprint sensors are single-channel grayscale images. Indicates the number of columns of the image (the width of the image), Indicates the number of rows in the image (the height of the image).

[0054] A fingerprint image matching and recognition method may include the following steps:

[0055] Step S1: input a small area fingerprint image Gaussian filtering is used to suppress noise and obtain an image .

[0056] Specifically, is a Gaussian kernel, then:

[0057] ,

[0058] ,

[0059] Among them, the symbol " " indicates convolution operation, Gaussian filter kernel parameters is 2.

[0060] Step S2, calculate image Sobel gradient , .

[0061] Specifically, the calculation formula is as follows:

[0062] .

[0063] Step S3, calculate the amplitude .

[0064] Specifically, the calculation formula is as follows:

[0065] .

[0066] Further, Perform mean filtering with a radius of 7 to obtain , then for each pixel position Perform background judgment, if If it is less than the preset threshold, It is marked as background, otherwise it is marked as foreground.

[0067] The selection of the threshold is generally related to the semiconductor fingerprint acquisition sensor, and is obtained by statistically analyzing its imaging distribution. The thresholds used by fingerprint acquisition sensors produced by different manufacturers are generally different.

[0068] Step S4: for each foreground pixel position , calculate the candidate feature point check vector , .

[0069] Specifically, the calculation formula is as follows:

[0070] ,

[0071] Furthermore, if If the number of components greater than or equal to 0 in is greater than or equal to 3, then the pixel position is marked. is a "maximum" candidate point; if If the number of components less than or equal to 0 in is greater than or equal to 3, then mark the pixel position is a candidate point for "extreme minimum".

[0072] Step S5: Check the relative positions of all candidate points. If two "maximum" candidate points are found, and If the 8-neighborhood adjacent condition is met, the neighborhood differences are calculated respectively. , .

[0073] Specifically, for a certain pixel , its neighborhood difference The calculation formula is:

[0074] ;

[0075] Furthermore, if , then delete The "maximum" candidate point of the point is marked, otherwise it is deleted The "maximum" candidate point of the point is marked. Similarly, if two "minimum" candidate points are found and If the 8-neighborhood adjacent condition is met, the neighborhood differences are calculated respectively. , .like , then delete The "minimum" candidate point of the point is marked, otherwise it is deleted The “minimum” candidate point of the point is marked, and the point left with the mark is the detected feature point.

[0076] The detected Feature points , is the number of feature points detected in step S5, calculated using the circular angle voting method According to the symmetry characteristics of periodic lines of fingerprint images, the non-polarity direction of each feature point is calculated.

[0077] Among them, the non-polarity direction refers to the original normal range of The angle of the forward and reverse directions is merged into one direction, and the angle range becomes The reason why the non-polarity direction is calculated is that the fingerprint has a high local symmetry. In many cases, the calculation of the polarity direction will have a large deviation, and the main direction is not significant, resulting in a large deviation in the feature descriptor extracted subsequently, and the match cannot be made. If the auxiliary direction is used to make up for the deficiency, it will bring a lot of storage and calculation overhead. Therefore, the present invention aims at this difficulty and proposes to first calculate the non-polarity direction with high accuracy, and then carefully design and extract a high-discrimination feature that is independent of the direction polarity, thereby effectively overcoming this difficulty.

[0078] Step S6, calculate the feature points is the center and the radius is The non-polarity direction of all pixels within the circle.

[0079] Specifically, for the pixel position The non-polar direction The definition is as follows:

[0080] ;

[0081] in, represents the imaginary unit, It means taking the principal value of the complex number’s argument. .

[0082] Step S7, obtaining After finding the non-polar direction of each pixel in the surrounding circle, these values ​​are statistically analyzed with weighted histograms. .

[0083] Specifically, the statistical histogram is divided into columns with every 5 degrees of direction, that is, For the first column, This is the second column, and so on, for a total of 36 columns. Each pixel position in the surrounding circle The direction weight is the amplitude of the point , let the statistical histogram be , then for the first The statistics for the columns are:

[0084] ;

[0085] in, is the Gaussian rounding function.

[0086] Step S8, histogram Smoothing process :

[0087] ;

[0088] in, , , , , .

[0089] Step S9, using a fitting method to estimate the precise non-polarity direction, in the histogram Find the largest bar in the Column, then the feature point Directions for:

[0090] ;

[0091] in, , , Represents the modulo operation.

[0092] Since the direction of feature points has no polarity, it is necessary to design polarity-independent features. In the feature point , with its coordinates As the origin, with direction for A rectangular coordinate system is established in the positive direction of the axis. In this coordinate system, a coordinate system with the origin as the center is constructed with a size of The square area , .

[0093] Step S10, by bilinear interpolation from the image Cut out a square area , treat it as a small image block, pay attention to the feature points here The position has been changed to a small image block The center of Middle Line The gray value at the column is express, The row and column indexes start from 0.

[0094] Step S11, using the down-sampling mirror difference method to calculate the image block The 128-dimensional centrosymmetric eigenvector of , its The calculation formula of the dimensional component is as follows ( ):

[0095] ,

[0096] Step S12, use The Hadamard matrix Action eigenvector get .

[0097] in, .

[0098] Specifically, the 128-order Hadamard matrix The construction method is as follows:

[0099] ,

[0100] ,

[0101] Among them, the symbol " ” is the Kronecker product.

[0102] Step S13, using a symbol quantization method to Hash to 128-dimensional Hamming space to get the final hash feature .

[0103] Specifically, the calculation formula is as follows:

[0104] .

[0105] in, The first row is all 1, and All components of are non-negative, so Always 1, which is an invalid bit. In order to make full use of the stored information, the present invention will The value of is changed to the "maximum" and "minimum" attributes of the feature point obtained in the detection stage. If the feature point is marked as a "maximum" candidate point, then ,otherwise .

[0106] Assume that the two small-area fingerprint images to be matched are and , the number of feature points obtained by feature extraction is and , The obtained features are , The obtained features are , and The image size is , Indicates the number of columns of the image (the width of the image), Indicates the number of rows in the image (the height of the image).

[0107] Step S14: Each feature point in , calculate its Each feature point The Hamming similarity .

[0108] Specifically, the calculation formula is as follows:

[0109] ,

[0110] Among them, the symbol " " indicates XOR operation, similarity The closer to 1, the and The higher the similarity, the higher the probability of matching. The closer to 0, the and The lower the similarity, the lower the probability of a match.

[0111] Step S15: Sort from largest to smallest and take the top three as The candidate pairing points of .

[0112] Through the above process, Each feature point in , have found their The three most likely matching feature points are , , .

[0113] The present invention has been proved through a large number of experiments that the correspondence between two small-area fingerprint images that are truly matched is generally a similarity transformation. and is a corresponding point, then:

[0114] ,

[0115] in, is the similarity ratio, is the rotation angle, , is the translation parameter.

[0116] The present invention adopts the method of random clustering to carry out and The best consistent estimate between .

[0117] Step S16, initializing the parameter space array ,in , , , Initialized to 0.

[0118] Specifically, randomly select Two feature points in and , through the relationship between suspected matching points and Calculate the parameters of the similarity transformation relationship , , , ,here and It can take values ​​1, 2, or 3, so there are There are 9 situations in total.

[0119] ,

[0120] ,

[0121] ,

[0122] ,

[0123] if Not meeting the conditions: , then skip this case.

[0124] if Not meeting the conditions: , Then skip this case.

[0125] If satisfied, then in the parameter space array The weights are accumulated at the corresponding positions in:

[0126] ,

[0127] Every update Then, record the maximum value and the position of the maximum value.

[0128] Repeat step b to achieve times or The maximum value in exceeds 7.

[0129] Use the estimated optimal transformation parameters Generated from arrive Rigid transformation of

[0130] ,

[0131] calculate All feature points in exist The following image:

[0132] ,

[0133] if Candidate pairing points , , There is a point satisfy:

[0134] ,

[0135] ,

[0136] Then mark The points that match.

[0137] After the above process, we can obtain and The final matched feature points are judged when the number of matched feature points exceeds a certain threshold. Match , otherwise judge and No match.

[0138] Through the above process, the embodiment of the present invention extracts the origin symmetric hash features in the fingerprint image and designs an efficient feature matching algorithm, which not only takes into account the local details of the fingerprint image, but also incorporates the global symmetry information, making the feature description more comprehensive and accurate, and improving the accuracy and robustness of the matching. It realizes the rapid and accurate recognition of small-area fingerprint images, which not only reduces the amount of calculation, but also improves the recognition speed and matching accuracy, thereby effectively solving the problem that the prior art is easily affected by noise and acquisition conditions, and has low recognition efficiency and low matching accuracy.

[0139] The following is an embodiment of the device of the present invention, which can be used to execute the fingerprint image matching and recognition method involved in the present invention. For details not disclosed in the embodiment of the device of the present invention, please refer to the method embodiment of the fingerprint image matching and recognition method involved in the present invention.

[0140] See also Figure 6 , an embodiment of the present invention provides a fingerprint image matching and recognition device 800.

[0141] The fingerprint image matching and recognition device 800 includes but is not limited to: a feature point acquisition module 810 , a feature extraction module 830 , a similarity calculation module 850 and an image matching module 870 .

[0142] The feature point acquisition module 810 is used to acquire a fingerprint image to be matched, perform position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, and obtain feature points according to the relative positions of the candidate points.

[0143] The feature extraction module 830 is used to calculate the non-polarized direction of the feature point by using the circular domain angle voting method, and obtain the hash feature of the feature point by downsampling mirror difference, hash mapping and symbol quantization.

[0144] The similarity calculation module 850 is used to select feature points that meet the set conditions as candidate pairing points by calculating the Hamming similarity between each feature point in the fingerprint image to be matched and sorting them.

[0145] The image matching module 870 is used to calculate the matching transformation relationship between candidate pairing points through a random clustering strategy and verify the number of pairing points. If the number of pairing points meets the preset conditions, the fingerprint image to be matched is matched successfully.

[0146] It should be noted that, in the fingerprint image matching and recognition provided in the above embodiments, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the fingerprint image matching and recognition device will be divided into different functional modules to complete all or part of the functions described above.

[0147] In addition, the fingerprint image matching and recognition device and the fingerprint image matching and recognition method provided in the above embodiments belong to the same concept, and the specific manner in which each module performs operations has been described in detail in the method embodiments and will not be repeated here.

[0148] Figure 7 The present invention schematically illustrates the structure of an electronic device according to an exemplary embodiment.

[0149] It should be noted that the electronic device is only an example adapted to the present invention and cannot be considered to provide any limitation on the scope of use of the present invention. The electronic device cannot be interpreted as requiring dependence on or having Figure 7 One or more components of exemplary electronic device 2000 are shown.

[0150] The hardware structure of the electronic device 2000 may vary greatly due to differences in configuration or performance, such as Figure 7 As shown, the electronic device 2000 includes: a power supply 210 , an interface 230 , at least one memory 250 , and at least one central processing unit (CPU) 270 .

[0151] Specifically, the power supply 210 is used to provide operating voltage for each hardware device on the electronic device 2000 .

[0152] The interface 230 includes at least one wired or wireless network interface 231 for interacting with external devices. Of course, in other examples adapted by the present invention, the interface 230 may further include at least one serial-to-parallel conversion interface 233, at least one input-output interface 235, and at least one USB interface 237, etc. Figure 7 As shown, this is not intended to be a specific limitation.

[0153] The memory 250 is a carrier for storing resources, which may be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon include an operating system 251, an application 253 and data 255, etc. The storage method may be temporary storage or permanent storage.

[0154] Among them, the operating system 251 is used to manage and control various hardware devices and application programs 253 on the electronic device 2000 to enable the central processor 270 to calculate and process the massive data 255 in the memory 250. It can be WindowsServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0155] The application program 253 is a computer-readable instruction that performs at least one specific task based on the operating system 251, and may include at least one module ( Figure 7 (not shown), each module may respectively contain computer-readable instructions for the electronic device 2000. For example, the fingerprint image matching and recognition device may be regarded as an application 253 deployed in the electronic device 2000.

[0156] The data 255 may be signal information, etc., stored in the memory 250 .

[0157] The central processor 270 may include one or more processors and is configured to communicate with the memory 250 through at least one communication bus to read the computer-readable instructions stored in the memory 250, thereby realizing the operation and processing of the mass data 255 in the memory 250. For example, the fingerprint image matching recognition method is completed by the central processor 270 reading a series of computer-readable instructions stored in the memory 250.

[0158] In addition, the present invention can also be implemented by hardware circuits or hardware circuits combined with software. Therefore, the implementation of the present invention is not limited to any specific hardware circuits, software, or a combination of the two.

[0159] See also Figure 8 In an embodiment of the present invention, an electronic device 4000 is provided. The electronic device 4000 may include: a desktop computer, a laptop computer, a server, etc. with sensor recognition capability.

[0160] exist Figure 8 In the embodiment, the electronic device 4000 includes at least one processor 4001 and at least one memory 4003.

[0161] The data interaction between the processor 4001 and the memory 4003 can be realized through at least one communication bus 4002. The communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 4003. The communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0162] Optionally, the electronic device 4000 may further include a transceiver 4004, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception, etc. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0163] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0164] The memory 4003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program instructions or codes in the form of instructions or data structures and can be accessed by the electronic device 4000, but is not limited to this.

[0165] Computer-readable instructions are stored in the memory 4003 , and the processor 4001 can read the computer-readable instructions stored in the memory 4003 through the communication bus 4002 .

[0166] The computer-readable instructions are executed by one or more processors 4001 to implement the fingerprint image matching and recognition methods in the above-mentioned embodiments.

[0167] In addition, an embodiment of the present invention provides a storage medium on which computer-readable instructions are stored. The computer-readable instructions are executed by one or more processors to implement the fingerprint image matching and recognition method as described above.

[0168] A computer program product is provided in an embodiment of the present invention. The computer program product includes computer-readable instructions. The computer-readable instructions are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the fingerprint image matching and recognition method as described above.

[0169] Compared with the related art, the present invention has the following beneficial effects:

[0170] 1. The present invention obtains the fingerprint image to be matched first, performs position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, obtains feature points according to the relative positions of the candidate points, uses the circular domain angle voting method to calculate the non-polarity direction of the feature points, obtains the hash features of the feature points by downsampling mirror difference, hash mapping and symbol quantization, calculates the Hamming similarity between each feature point in the feature points of the fingerprint image to be matched and sorts them, selects the feature points that meet the set conditions as candidate pairing points, calculates the matching transformation relationship between the candidate pairing points by random clustering strategy and verifies the number of pairing points, and successfully matches the fingerprint image to be matched if the number of pairing points meets the preset conditions, extracts the origin symmetric hash features in the fingerprint image and designs an efficient feature matching algorithm, not only considers the local details of the fingerprint image, but also integrates the global symmetry information, so that the feature description is more comprehensive and accurate, and the matching accuracy and robustness are improved. The invention realizes the rapid and accurate recognition of small-area fingerprint images, not only reduces the amount of calculation, but also improves the recognition speed and matching accuracy, thereby effectively solving the problem that the prior art is easily affected by noise and acquisition conditions, has low recognition efficiency and low matching accuracy.

[0171] 2. The present invention can improve recognition efficiency: By designing a feature extraction algorithm based on origin-symmetric hash features, the present solution can efficiently extract representative and distinguishing feature information from a small-area fingerprint image. This efficient feature extraction method reduces the amount of data to be processed later, thereby speeding up recognition.

[0172] 3. The present invention can enhance the matching accuracy: the origin symmetric hash feature not only considers the local details of the fingerprint image, but also incorporates the global symmetry information, making the feature description more comprehensive and accurate. This feature description method can better resist noise interference and image quality changes during the matching process, and improves the accuracy and robustness of the matching.

[0173] 4. The present invention can reduce computational complexity: Compared with traditional complex image processing algorithms, the hash feature mapping and quantization method adopted in this solution greatly simplifies the computational process. By mapping the hash function, the high-dimensional feature vector is converted into a low-dimensional hash code, which not only reduces the storage space, but also speeds up the similarity calculation in the matching process.

[0174] 5. The present invention can expand application scenarios: Since the present solution is optimized for small-area fingerprint images, it is particularly suitable for scenarios with strict space requirements or high portability requirements, such as smart phones, wearable devices, smart cards, etc. These devices are usually limited by size and power consumption and require fast and accurate identity authentication solutions, and this solution just meets this demand.

[0175] 6. The present invention can improve user experience: Fast and accurate fingerprint recognition can significantly improve the user experience. Users no longer need to wait for a long verification process, nor worry about recognition failures due to poor image quality. This smooth and reliable identity authentication method will enhance user trust and satisfaction with the product.

[0176] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0177] The above descriptions are only partial embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A fingerprint image matching and recognition method, characterized in that: The method comprises: Acquire a fingerprint image to be matched, perform position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, and obtain feature points according to the relative positions of the candidate points; Calculate all the pixels in the fingerprint image to be matched within a circle with a radius of r and the feature point as the center Non-polar direction , the calculation formula is as follows: ； in, and represents the Sobel gradient of the fingerprint image to be matched after Gaussian filtering, represents the imaginary unit, ; For all the pixels Non-polar direction Performing histogram statistics and smoothing processing to obtain a histogram, and fitting the histogram to obtain the non-polarity direction of the feature point; Calculate a square feature domain of each feature point according to the non-polarity direction of the feature point, calculate a feature vector in the square feature domain using a mirror difference method, perform Hadamard hash mapping on the feature vector and quantize it to obtain a hash feature; Obtain two sets of feature point sets according to the position, direction and hash feature of each of the feature points, calculate the hash feature Hamming distance between the two sets of feature point sets and sort them, and select the feature points that meet the set conditions as candidate pairing points; The matching transformation relationship between the candidate pairing points is calculated by random clustering strategy and the number of pairing points is verified. If the number of pairing points meets the preset conditions, the fingerprint image to be matched is matched successfully.

2. The fingerprint image matching and recognition method according to claim 1, characterized in that: The step of performing position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, and obtaining feature points according to the relative positions of the candidate points, includes: Using Gaussian filtering to suppress noise on the fingerprint image to be matched, and calculating the Sobel gradient and gradient amplitude of the fingerprint image to be matched; Performing foreground and background segmentation on the fingerprint image to be matched by using the gradient amplitude and a preset threshold to obtain a fingerprint foreground area, and calculating a check vector for each pixel point in the fingerprint foreground area; A candidate point is obtained according to the check vector of the pixel point, and the candidate point is screened by neighborhood difference and 8-neighborhood adjacent conditions to obtain a feature point.

3. The fingerprint image matching and recognition method according to claim 2, characterized in that: The step of performing foreground and background segmentation on the fingerprint image to be matched by using the gradient amplitude and a preset threshold to obtain a fingerprint foreground area includes: Performing mean filtering with a radius of 7 on the gradient amplitude to obtain the amplitude mean of each pixel point of the fingerprint image to be matched; If the amplitude mean of the pixel point is not less than a set threshold, it is marked as a fingerprint foreground area; the set threshold is obtained through the imaging distribution of the acquisition sensor of the fingerprint image to be matched.

4. The fingerprint image matching and recognition method according to claim 2, characterized in that: The step of obtaining candidate points according to the check vector of the pixel point, and screening the candidate points by neighborhood difference and 8-neighborhood adjacent conditions to obtain feature points includes: If the number of components greater than or equal to 0 in the check vector of the pixel point is greater than or equal to 3, the pixel point is marked as a maximum candidate point; If the number of components less than or equal to 0 in the check vector of the pixel point is greater than or equal to 3, the pixel point is marked as a minimum candidate point; If the two minimum candidate points satisfy the 8-neighborhood adjacent condition, the neighborhood difference of the two minimum candidate points is calculated, and the minimum candidate point with the smallest neighborhood difference is taken as the feature point; If the two maximum candidate points satisfy the 8-neighborhood adjacent condition, the neighborhood difference of the two maximum candidate points is calculated, and the maximum candidate point with the largest neighborhood difference is taken as the feature point.

5. The fingerprint image matching and recognition method according to claim 1, characterized in that: The method of calculating a square feature domain of each feature point according to the non-polarity direction of the feature point, calculating a feature vector in the square feature domain using a mirror difference method, and performing Hadamard hash mapping on the feature vector and quantizing the feature vector to obtain a hash feature includes: According to the non-polarity direction of the feature point, a square feature domain is constructed with the feature point as the center, and a down-sampling mirror difference method is used to calculate the central symmetric feature vector of the square feature domain; The feature vector is hash mapped using a Hadamard matrix, and the hash feature of the feature point is obtained by symbolic quantization.

6. The fingerprint image matching and recognition method according to claim 1, characterized in that: The step of calculating the matching transformation relationship between the candidate pairing points by a random clustering strategy and verifying the number of pairing points includes: By means of a random clustering strategy and setting matching point pair relationships, the transformation parameters of the similarity transformation relationship between the candidate paired points are calculated; the transformation parameters include a similarity ratio, a rotation angle, and a translation parameter; A consistency matching transformation is performed according to the transformation parameters, and the matching of the candidate pairing points is verified to obtain the number of pairing points.

7. A fingerprint image matching and recognition device, characterized in that: The device comprises: A feature point acquisition module is used to acquire a fingerprint image to be matched, perform position preprocessing and candidate point verification on the fingerprint image to be matched to obtain candidate points, and obtain feature points according to the relative positions of the candidate points; The feature extraction module is used to calculate all the pixels in the fingerprint image to be matched within a circle with a radius of r and the feature point as the center. Non-polar direction , the calculation formula is as follows: ； in, and represents the Sobel gradient of the fingerprint image to be matched after Gaussian filtering, represents the imaginary unit, ; For all the pixels Non-polar direction Performing histogram statistics and smoothing processing to obtain a histogram, and fitting the histogram to obtain the non-polarity direction of the feature point; Calculate a square feature domain of each feature point according to the non-polarity direction of the feature point, calculate a feature vector in the square feature domain using a mirror difference method, perform Hadamard hash mapping on the feature vector and quantize it to obtain a hash feature; A similarity calculation module is used to obtain two sets of feature point sets according to the position, direction and hash feature of each feature point, calculate the hash feature Hamming distance between the two sets of feature point sets and sort them, and select the feature points that meet the set conditions as candidate pairing points; The image matching module is used to calculate the matching transformation relationship between the candidate pairing points through a random clustering strategy and verify the number of pairing points. If the number of pairing points meets the preset conditions, the fingerprint image to be matched is matched successfully.

8. An electronic device, characterized in that: include: at least one processor and at least one memory, wherein: The memory has computer-readable instructions stored thereon; The computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the fingerprint image matching and recognition method according to any one of claims 1 to 6.

9. A storage medium having computer-readable instructions stored thereon, characterized in that: The computer-readable instructions are executed by one or more processors to implement the fingerprint image matching and recognition method according to any one of claims 1 to 6.

Images