A biometric recognition method and system

Abstract

The present invention belongs to the field of biometric identification, and discloses a biometric identification method and system. The method includes: S1, calculating the pre-recorded first images containing biometric information stored in the database, dividing each first image into multiple first comparison regions respectively, calculating the first comparison coefficients of each first comparison region, and storing all the first comparison regions and the first comparison coefficients; S2, obtaining a second image containing the biometric information of the person whose biometric identification needs to be performed; S3, comparing the second image with each first image in turn to determine whether there is a first image in the database that meets the identity verification standard. If so, it means that the biometric identification is passed. The present invention preferentially obtains the dissimilarity of the first comparison regions with more feature information, dynamically introduces the distribution information of the first comparison regions with the similarity meeting the requirements, and can obtain the judgment result through fewer comparison times.

Description

Technical Field

[0001] The present invention relates to the field of biometric identification, and particularly to a biometric identification method and system. Background Art

[0002] Biometric identification refers to obtaining images containing biometric features of a human body, such as face images, fingerprint images, pupil images, etc., then extracting features from these images, and comparing the extracted features with the features of the same type of images pre-recorded (for example, calculating the similarity), and judging whether the biometric identification is passed according to the comparison result.

[0003] In the prior art, usually the features of the whole image are obtained first, and then the comparison is made based on the obtained features. If the number of pre-recorded images is too large, the speed of biometric identification will be too slow. In order to improve the identification efficiency, in the prior art, there has also appeared a method of first dividing into multiple regions of the same size and then making comparisons based on the regions. The advantage of this method is that it is not necessary to calculate all pixel points completely to obtain features. Each time, only the features of a part of the region need to be calculated. When many regions do not pass the feature comparison, it means that it is no longer necessary to compare the remaining regions, thereby further improving the efficiency of feature comparison. However, this method of partition comparison also has certain defects, that is, since there is no restriction on the comparison order of the regions, if the regions compared first are all regions with less feature information, it will also take a long time to obtain the comparison result. Summary of the Invention

[0004] The purpose of the present invention is to disclose a biometric identification method and system to solve the technical problems proposed in the background art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] The present invention provides a biometric identification method, including:

[0007] S1, calculating the first images containing biometric information pre-recorded in the database, dividing each first image into multiple first comparison regions respectively, calculating the first comparison coefficients of each first comparison region respectively, and storing all the first comparison regions and the first comparison coefficients;

[0008] S2, obtaining a second image containing the biometric information of the person to be biometrically identified;

[0009] S3, comparing the second image with each first image in turn, and judging whether there is a first image in the database that meets the identity verification standard. If so, it means that the biometric identification is passed, and the comparison ends;

[0010] Among them, for the first image P1, it is judged whether P1 meets the authentication standard in the following way:

[0011] S30: Represent the set of the first comparison regions corresponding to P1 with UP1, obtain the first comparison region ARE1 with the largest first comparison coefficient in UP1, and store ARE1 in the set UP2;

[0012] S31: Obtain the dissimilarity score of ARE1, and delete ARE1 from UP1;

[0013] S32: Judge whether the sum of the dissimilarity scores of the first comparison regions in UP2 is greater than the set score threshold. If so, it means that P1 does not meet the authentication standard. If not, go to S33;

[0014] S33: Calculate the second comparison coefficient of each first comparison region in UP1 based on the first comparison coefficient and UP2 respectively;

[0015] S34: Obtain the dissimilarity score of the first comparison region with the largest second comparison coefficient in UP1, delete the first comparison region with the largest second comparison coefficient from UP1, and store the first comparison region with the largest second comparison coefficient in UP2;

[0016] S35: Judge whether the number of the remaining first comparison regions in UP1 is less than 1. If so, go to S32. If not, it means that P1 meets the authentication standard.

[0017] Preferably, the first image is divided into multiple first comparison regions, including:

[0018] Use a preprocessing algorithm to process the first image to obtain the image PF;

[0019] Use NL and NW to represent the length and width of PF respectively;

[0020] Divide PF into image regions with a length of and a width of , where N represents a preset integer;

[0021] Perform an enlargement process on the obtained image regions to obtain multiple first comparison regions.

[0022] Preferably, performing an enlargement process on the obtained image regions to obtain multiple first comparison regions, including:

[0023] S41: Take all the image regions as elements in the set ARU;

[0024] S42. Randomly extract an image region ru from the ARU, perform an enlargement process on ru to obtain a first comparison region, and store the obtained first comparison region in the set CMPU;

[0025] S43. Determine whether there is still an image region in the ARU. If so, enter the second step; if not, use the first comparison region in the CMPU as the determined first comparison region.

[0026] Preferably, performing an enlargement process on ru to obtain a first comparison region includes:

[0027] S51. Use ru as the comparison target and store ru in the set FA;

[0028] S52. Obtain the set ruA of all image regions adjacent to the comparison target in the ARU;

[0029] S53. Calculate the competition value of each image region in ruA respectively;

[0030] S54. Determine whether the maximum value among all the competition values is greater than the set competition value threshold. If so, store the image region with the largest competition value in ruA in the set FA and enter S55; if not, stop the processing and splice all the image regions in FA to obtain the first comparison region;

[0031] S55. Delete the image regions in FA from the ARU;

[0032] S56. Use the image region with the largest competition value in ruA as the updated comparison target and enter S52.

[0033] Preferably, using a preprocessing algorithm to process the first image to obtain an image PF includes:

[0034] Perform noise reduction processing on the first image to obtain an intermediate image;

[0035] Perform enhancement processing on the intermediate image to obtain the image PF.

[0036] Preferably, the calculation formula of the first comparison coefficient is:

[0037]

[0038] is the first comparison coefficient of the first comparison region q, is the number of pixel points belonging to the set EG existing in the first comparison region q, The number of pixel points in the first comparison region q, represents the distance between the coordinates of the center of the first comparison region q and the coordinates of the center of the first image, Represents the length of the first image, and qu is the set of pixel points in the first comparison region q. Is the image gradient of pixel point z. Represents the maximum value of the image gradients of all pixel points in the first comparison regions, and d1, d2, and d3 are the first weighting factor, the second weighting factor, and the third weighting factor respectively.

[0039] Preferably, the process of obtaining EG is as follows:

[0040] Calculate the detection values in the horizontal direction and the vertical direction of each pixel point in the first comparison region q respectively.

[0041] Use the following formula to calculate the comparison value of each pixel point in the first comparison region q respectively:

[0042]

[0043] Represents the comparison value of pixel point c. And Represent the detection value in the horizontal direction and the detection value in the vertical direction of pixel point c respectively.

[0044] Take the pixel points with comparison values greater than the set comparison value threshold as the elements of EG.

[0045] Preferably, for the first comparison region im, the process of obtaining the dissimilarity score of im includes:

[0046] Store the coordinates of all pixel points in im into the set LKim.

[0047] Obtain the second comparison region formed by the pixel points corresponding to all the coordinates in LKim in the second image PS.

[0048] Calculate the similarity between im and the second comparison region.

[0049] Judge whether the similarity is greater than the pre-set similarity threshold. If not, then calculate the dissimilarity score of im using the following formula :

[0050]

[0051] Is the first comparison coefficient of im, and CMPU1 is the set of the first comparison regions in P1. Represents the first comparison coefficient of the first comparison region i.

[0052] If so, the dissimilarity score of im is 0.

[0053] Preferably, calculating a second comparison coefficient for each first comparison region in UP1 based on the first comparison coefficient and UP2 respectively includes:

[0054] Calculating the second comparison coefficient using the following formula:

[0055]

[0056] is the second comparison coefficient of the first comparison region j, is the first comparison coefficient of the first comparison region j, represents the number of first comparison regions belonging to UP2 among all first comparison regions adjacent to the first comparison region j, and NUP2 represents the total number of first comparison regions in UP2, represents the fourth weighting factor.

[0057] The present invention also provides a biometric recognition system, including a preprocessing module, an acquisition module, and a comparison module;

[0058] The preprocessing module is used to calculate for the first images containing biometric information pre-recorded in the database, divide each first image into multiple first comparison regions respectively, calculate the first comparison coefficient of each first comparison region respectively, and store all the first comparison regions and the first comparison coefficients;

[0059] The acquisition module is used to acquire a second image containing the biometric information of the person to be biometrically recognized;

[0060] The comparison module is used to compare the second image with each first image in turn, determine whether there is a first image that meets the identity verification standard in the database. If so, it means that the biometric recognition is passed and the comparison ends;

[0061] Among them, for the first image P1, it is judged whether P1 meets the identity verification standard through the following method:

[0062] S30, use UP1 to represent the set of first comparison regions corresponding to P1, obtain the first comparison region ARE1 with the largest first comparison coefficient in UP1, and store ARE1 in the set UP2;

[0063] S31, obtain the dissimilarity score of ARE1, and delete ARE1 from UP1;

[0064] S32, judge whether the sum of the dissimilarity scores of the first comparison regions in UP2 is greater than the set score threshold. If so, it means that P1 does not meet the identity verification standard. If not, enter S33;

[0065] S33. Calculate the second comparison coefficient for each first comparison region in UP1 based on the first comparison coefficient and UP2 respectively;

[0066] S34. Obtain the dissimilarity score of the first comparison region with the largest second comparison coefficient in UP1, delete the first comparison region with the largest second comparison coefficient from UP1, and store the first comparison region with the largest second comparison coefficient in UP2;

[0067] S35. Determine whether the number of remaining first comparison regions in UP1 is less than 1. If so, enter S32. If not, it means that P1 meets the identity verification standard.

[0068] Beneficial effects:

[0069] In the present invention, by pre-calculating the first comparison region and the first comparison coefficient, and then in the process of judging whether there is a first image that meets the identity verification standard in the second judgment database, by first calculating the dissimilarity of the first comparison region with the largest first comparison coefficient, and then based on the first comparison coefficient, circularly obtain the second comparison coefficient for the remaining first comparison regions whose dissimilarity has not been determined, and determine the comparison order through the second comparison coefficient, so that it is possible to preferentially obtain the dissimilarity of the first comparison region with more feature information. And the second comparison coefficient dynamically introduces the distribution information of the first comparison regions whose similarity meets the requirements during the calculation process, so that it is possible to more preferentially judge the first comparison regions that are more relevant to the already determined high-similarity regions and are more meaningful for judgment, which is beneficial to judging whether the first image meets the identity verification standard with fewer comparison times and can effectively improve the judgment efficiency. Brief description of the drawings

[0070] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0071] Figure 1 It is a schematic diagram of a biometric recognition method of the present invention. Detailed implementation manners

[0072] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0073] The present invention provides a biometric recognition method, including:

[0074] S1. Calculate the pre-recorded first images containing biometric information stored in the database. Divide each first image into multiple first comparison regions respectively, calculate the first comparison coefficients of each first comparison region respectively, and store all the first comparison regions and the first comparison coefficients.

[0075] The database of the present invention can be a local database or a cloud database. In some small or local biometric recognition systems, image data may be stored in a local database (such as a relational database or a non-relational database). These databases may be data stored on a server, a workstation or an embedded device. In a larger-scale or remotely accessible system, the pre-recorded images are usually stored in a cloud database. The cloud database can provide advantages such as higher scalability, fault tolerance and distributed storage. Common cloud service providers such as AWS, Google Cloud, Microsoft Azure, etc. all provide data storage solutions related to biometric recognition.

[0076] Authorized personnel pre-enter the first images containing biometric information, and the first images provide a basis for judgment in the subsequent biometric recognition process.

[0077] S2. Obtain a second image containing the biometric information of the person whose biometric needs to be recognized.

[0078] The first image and the second image of the present invention have the same image type, and the image type can be a face image, a fingerprint image, a pupil image, etc.

[0079] Furthermore, the first image and the second image of the present invention have the same resolution.

[0080] S3. Compare the second image with each first image in turn, and judge whether there is a first image that meets the identity verification standard in the database. If so, it means that the biometric recognition is passed and the comparison ends; if not, it means that the biometric recognition fails.

[0081] Among them, for the first image P1, judge whether P1 meets the identity verification standard through the following method:

[0082] S30. Use UP1 to represent the set of the first comparison regions corresponding to P1, obtain the first comparison region ARE1 with the largest first comparison coefficient in UP1, and store ARE1 in the set UP2.

[0083] S31. Obtain the non-similarity score of ARE1, and delete ARE1 from UP1.

[0084] S32. Determine whether the sum of the dissimilarity scores of the first comparison regions in UP2 is greater than a set score threshold. If so, it indicates that P1 does not meet the authentication standard. If not, proceed to S33;

[0085] S33. Calculate the second comparison coefficient of each first comparison region in UP1 based on the first comparison coefficient and UP2 respectively;

[0086] S34. Obtain the dissimilarity score of the first comparison region with the largest second comparison coefficient in UP1, delete the first comparison region with the largest second comparison coefficient from UP1, and store the first comparison region with the largest second comparison coefficient in UP2;

[0087] S35. Determine whether the number of remaining first comparison regions in UP1 is less than 1. If so, proceed to S32. If not, it indicates that P1 meets the authentication standard.

[0088] In the process of comparing the second image with the first image, the present invention calculates the dissimilarity score. When it is found that the dissimilarity score is greater than the set score threshold, the comparison of the current first image is stopped, and the comparison of the next first image is continued. In this way, it is possible to obtain the comparison result without having to compare all pixel points, which can effectively improve the comparison efficiency.

[0089] The score threshold of the present invention can be a value between 0.2 and 0.3, such as 0.25. The value of the score threshold can be adjusted according to security requirements. The higher the security requirements, the lower the score threshold is set.

[0090] In the process of performing regional comparison, the present invention does not determine the comparison order based on a static first comparison coefficient, because this cannot utilize the distribution information of the image details in the second image and the position information of the regions with high similarity. As a result, when determining the comparison order based on the first comparison coefficient, after comparing the first comparison regions with a relatively early comparison order, due to the large positional gap between the first comparison regions, it may not be possible to perform continuous comparison on a relatively large local area well, resulting in a relatively low representativeness of the comparison result, that is, more first comparison regions need to be compared to determine whether it meets the authentication standard. After the present invention introduces dynamic position data, it is possible to give priority to comparing a relatively large local area with a relatively high first comparison coefficient. After comparing multiple first comparison regions with a relatively small area, it is possible to simultaneously perform comparison on a relatively large local area composed of multiple first comparison regions with close positions. This makes the comparison results with a relatively early comparison order more representative, and it is possible to obtain the comparison result with fewer comparison times. Preferably, the first image is divided into multiple first comparison regions, including:

[0091] Process the first image using a preprocessing algorithm to obtain image PF;

[0092] Use NL and NW to represent the length and width of PF respectively;

[0093] Divide PF into image regions with a length of and a width of , where N represents a preset integer;

[0094] Perform an enlargement process on the obtained image regions to obtain multiple first comparison regions.

[0095] The preset integer in the present invention can be 50 or 100, or a positive integer with other values, which can be adjusted according to needs. The higher the resolution of the image, the larger the value of the preset integer is generally set.

[0096] Processing using a preprocessing algorithm can improve the recognition accuracy in subsequent processing stages.

[0097] Preferably, performing an enlargement process on the obtained image regions to obtain multiple first comparison regions includes:

[0098] S41, Take all the image regions as elements in the set ARU;

[0099] S42, Randomly extract an image region ru from ARU, perform an enlargement process on ru to obtain a first comparison region, and store the obtained first comparison region in the set CMPU;

[0100] S43, Determine whether there are still image regions in ARU. If so, enter the second step. If not, use the first comparison regions in CMPU as the determined first comparison regions.

[0101] The above process continuously extracts an image region ru from ARU for enlargement processing, and the image region actually pointed to by ru continuously changes until there are no image regions in ARU to end the processing process, thereby obtaining multiple first comparison regions.

[0102] Preferably, performing an enlargement process on ru to obtain a first comparison region includes:

[0103] S51, Take ru as the comparison target and store ru in the set FA;

[0104] S52, Obtain the set ruA of all image regions adjacent to the comparison target in ARU;

[0105] S53, Calculate the competition value of each image region in ruA respectively;

[0106] S54. Determine whether the maximum value among all the competition values is greater than the set competition value threshold. If so, store the image region with the maximum competition value in ruA into the set FA, and proceed to S55; if not, stop the processing, splice all the image regions in FA to obtain the first comparison region.

[0107] S55. Delete the image regions in FA from ARU.

[0108] S56. Use the image region with the maximum competition value in ruA as the updated comparison target, and proceed to S52.

[0109] In the enlargement process of the present invention, the image region adjacent to the position of the comparison target in ruA and having the maximum competition value is incorporated into the range of the first comparison region where the comparison target is located, realizing the self-adaptive enlargement of the area of the first comparison region, and enabling as many image regions originally belonging to the same region as possible to be placed in the same first comparison region. In this way, the obtained first comparison region has an irregular shape, but can incorporate a local region (such as the mouth region) as completely as possible. When performing sub-region comparison with the second image, it is beneficial to improve the integrity of the image features obtained during the comparison and obtain a more accurate comparison result.

[0110] Preferably, the calculation formula for the competition value is:

[0111]

[0112] represents the competition value of the image region r in ruA, represents the edge judgment value, and respectively represent the number of pixel points with gray value equal to v in r and the comparison target, represents the edge judgment weight;

[0113] The calculation formula for the edge judgment value is:

[0114]

[0115] and are respectively the average abscissa and ordinate of the pixel points belonging to the image edge in the image region r; and are respectively the average abscissa and ordinate of the pixel points belonging to the image edge in the comparison target.

[0116] The competition value of the present invention is calculated from two directions: the edge judgment value and the difference in the number of pixel points with various gray values. If the edge judgment value is smaller and the difference in the number of pixel points with various gray values is smaller, it indicates that the probability that r and the comparison target belong to the same region is higher, thus achieving the adaptive acquisition of the first comparison region. The edge judgment value is obtained by calculating the difference between the average coordinates of the pixel points belonging to the image edge. The greater the difference, the smaller the correlation degree between the two image regions. Since not only the difference in the number of pixel points with different gray values is considered, but also the position difference of the edges in the image region is considered, the comprehensive difference between the two image regions can be represented more accurately.

[0117] Preferably, the edge judgment weight is .

[0118] Preferably, the competition value threshold is 0.7.

[0119] Preferably, a preprocessing algorithm is used to process the first image to obtain the image PF, including:

[0120] Perform noise reduction processing on the first image to obtain an indirect image;

[0121] Perform enhancement processing on the indirect image to obtain the image PF.

[0122] The noise reduction processing of the present invention can be implemented by algorithms such as median filtering. The enhancement processing can be implemented by histogram equalization. By adjusting the contrast of the image, the gray distribution of the image is made uniform, achieving the effect of enhancing the details of the image.

[0123] Preferably, the calculation formula of the first comparison coefficient is:

[0124]

[0125] is the first comparison coefficient of the first comparison region q, is the number of pixel points belonging to the set EG existing in the first comparison region q, The number of pixel points in the first comparison region q, represents the distance between the coordinate of the center of the first comparison region q and the coordinate of the center of the first image, represents the length of the first image, qu is the set of pixel points in the first comparison region q, is the image gradient of the pixel point z, represents the maximum value of the image gradients of all pixel points in the first comparison region, and d1, d2, and d3 are the first weighting factor, the second weighting factor, and the third weighting factor respectively.

[0126] The first comparison coefficient of the present invention is obtained based on static data, so it can be calculated and stored in advance, so that it can be directly called when used without having to calculate again during use, and the first comparison area for comparison can be determined more quickly, which is beneficial to improving the efficiency of biometric recognition.

[0127] The first comparison coefficient respectively considers the number of pixel points belonging to EG, the distance from the center of the first image, and the degree of dispersion of the image gradient. The more the number of pixel points belonging to EG, the smaller the distance from the center of the first image, and the greater the degree of dispersion of the image gradient, the greater the first comparison coefficient, indicating that the first comparison area contains more important image details, and the greater the impact on the result obtained after comparison. When comparing with the second image, this first comparison area should be given priority for comparison.

[0128] Preferably, the first weighting factor, the second weighting factor, and the third weighting factor are all 0.33.

[0129] Preferably, the process of obtaining EG is as follows:

[0130] Calculate the detection values in the horizontal direction and the vertical direction of each pixel point in the first comparison area q respectively;

[0131] Use the following formula to calculate the comparison value of each pixel point in the first comparison area q:

[0132]

[0133] represents the comparison value of pixel point c, and respectively represent the detection value in the horizontal direction and the detection value in the vertical direction of pixel point c;

[0134] Take the pixel points with comparison values greater than the set comparison value threshold as the elements of EG.

[0135] By calculating the detection values, pixel points with rich contour information can be quickly screened out as the elements in EG. This enables the calculation process of the first comparison coefficient to refer to pixel points with more contour information to improve the effectiveness of the first comparison coefficient in representing the importance of the first comparison area.

[0136] The calculation formula for the detection value in the horizontal direction is:

[0137]

[0138] x and y are the abscissa and ordinate of pixel point c respectively; is the gray value of the pixel point with abscissa x + 1 and ordinate y + 1;

[0139] The calculation formula for the detection value in the vertical direction is:

[0140] .

[0141] The set comparison value threshold can be 100.

[0142] Preferably, for the first comparison region im, the process of obtaining the dissimilarity score of im includes:

[0143] Store the coordinates of all pixel points in im into the set LKim;

[0144] Obtain the second comparison region formed by the pixel points corresponding to all the coordinates in LKim in the second image PS;

[0145] Calculate the similarity between im and the second comparison region;

[0146] Judge whether the similarity is greater than the pre-set similarity threshold. If not, calculate the dissimilarity score of im using the following formula :

[0147]

[0148] is the first comparison coefficient of im, CMPU1 is the set of the first comparison regions in P1, represents the first comparison coefficient of the first comparison region i;

[0149] If so, the dissimilarity score of im is 0.

[0150] The dissimilarity score of the present invention is represented by the ratio of the first comparison coefficient of im to the sum of the first comparison coefficients of all the first comparison regions. Therefore, if the first comparison coefficient of im is larger, when the similarity is less than the pre-set similarity threshold, the corresponding dissimilarity score of im is higher, so that when the difference between the first image and the second image is large, the comparison result can be obtained by only comparing some of the first comparison regions.

[0151] The similarity between im and the second comparison region can be calculated by the cosine similarity calculation algorithm. Correspondingly, the similarity threshold can be 0.8.

[0152] Preferably, calculate the second comparison coefficient of each first comparison region in UP1 based on the first comparison coefficient and UP2 respectively, including:

[0153] Calculate the second comparison coefficient using the following formula:

[0154]

[0155] is the second comparison coefficient for the first comparison region j. is the first comparison coefficient for the first comparison region j. represents the number of first comparison regions belonging to UP2 among all the first comparison regions adjacent to the first comparison region j, and NUP2 represents the total number of first comparison regions in UP2. represents the fourth weighting factor.

[0156] The second comparison coefficient of the present invention is obtained based on the first comparison coefficient. Since the first comparison coefficient can be directly called during calculation, the calculation speed is very fast. The second comparison coefficient introduces the dynamic gray value distribution related to the second image. When there are more other first comparison regions around j whose similarity with the local region in the second image meets the requirements, the larger the second comparison coefficient corresponding to j, indicating that the first comparison regions around j are more likely to belong to the overall local region. In this way, it is possible to give more priority to considering the comparison of a larger continuous range. While performing a small-range comparison, the comparison of a large range can be achieved as quickly as possible, which is beneficial to obtaining a more comprehensive comparison result and improving the accuracy of the comparison for determining whether the first image meets the requirements based on a small part of the first comparison regions.

[0157] The fourth weighting factor can be 0.8.

[0158] The present invention also provides a biometric recognition system, including a preprocessing module, an acquisition module, and a comparison module.

[0159] The preprocessing module is used to calculate the first images containing biometric information pre-entered in the database. Each first image is divided into multiple first comparison regions, and the first comparison coefficient of each first comparison region is calculated, and all the first comparison regions and the first comparison coefficients are stored.

[0160] The acquisition module is used to acquire a second image containing the biometric information of the person whose biometric recognition is required.

[0161] The comparison module is used to compare the second image with each first image in turn to determine whether there is a first image in the database that meets the identity verification standard. If so, it indicates that the biometric recognition is passed and the comparison ends.

[0162] Among them, for the first image P1, it is determined whether P1 meets the identity verification standard through the following method:

[0163] S30, use UP1 to represent the set of first comparison regions corresponding to P1, obtain the first comparison region ARE1 with the largest first comparison coefficient in UP1, and store ARE1 in the set UP2.

[0164] S31. Obtain the dissimilarity score of ARE1, and delete ARE1 from UP1;

[0165] S32. Determine whether the sum of the dissimilarity scores of the first comparison regions in UP2 is greater than a set score threshold. If so, it indicates that P1 does not meet the authentication standard. If not, proceed to S33;

[0166] S33. Calculate the second comparison coefficient of each first comparison region in UP1 based on the first comparison coefficient and UP2 respectively;

[0167] S34. Obtain the dissimilarity score of the first comparison region with the largest second comparison coefficient in UP1, delete the first comparison region with the largest second comparison coefficient from UP1, and store the first comparison region with the largest second comparison coefficient in UP2;

[0168] S35. Determine whether the number of remaining first comparison regions in UP1 is less than 1. If so, proceed to S32. If not, it indicates that P1 meets the authentication standard.

[0169] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art in the relevant technical field can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A biometric recognition method, characterized in that, Including: S1, calculate the pre - entered first images containing biometric information stored in the database. Divide each first image into multiple first comparison regions respectively, calculate the first comparison coefficients of each first comparison region respectively, and store all the first comparison regions and the first comparison coefficients. S2, obtain a second image containing the biometric information of the person whose biometric identification is required. S3, compare the second image with each first image in turn, and determine whether there is a first image in the database that meets the authentication standard. If so, it means that the biometric identification is passed and the comparison ends. Among them, for the first image P1, judge whether P1 meets the authentication standard through the following method: S30, use UP1 to represent the set of the first comparison regions corresponding to P1, obtain the first comparison region ARE1 with the largest first comparison coefficient in UP1, and store ARE1 in the set UP2. S31, obtain the dissimilarity score of ARE1, and delete ARE1 from UP1. S32, judge whether the sum of the dissimilarity scores of the first comparison regions in UP2 is greater than the set score threshold. If so, it means that P1 does not meet the authentication standard. If not, enter S33. S33, calculate the second comparison coefficients of each first comparison region in UP1 based on the first comparison coefficients and UP2 respectively. S34, obtain the dissimilarity score of the first comparison region with the largest second comparison coefficient in UP1, delete the first comparison region with the largest second comparison coefficient from UP1, and store the first comparison region with the largest second comparison coefficient in UP2. S35, judge whether the number of the remaining first comparison regions in UP1 is less than 1. If not, enter S32. If so, it means that P1 meets the authentication standard. The calculation process of the dissimilarity score is as follows: Obtain the second comparison region composed of the pixel points corresponding to all the pixel point coordinates of the first comparison region in the second image. Calculate the dissimilarity score between the first comparison region and the second comparison region. The calculation formula of the first comparison coefficient is: cmpfir q is the first comparison coefficient for the first comparison region q, neg q is the number of pixel points belonging to the set EG existing in the first comparison region q, np q is the number of pixel points in the first comparison region q, length q represents the distance between the coordinates of the center of the first comparison region q and the coordinates of the center of the first image, length max represents the length of the first image, qu is the set of pixel points in the first comparison region q, g z is the image gradient of the pixel point z, g max represents the maximum value of the image gradients of all pixel points in the first comparison regions, d1, d2, and d3 are the first weighting factor, the second weighting factor, and the third weighting factor respectively; The acquisition process of EG is as follows: Calculate the detection values in the horizontal direction and the vertical direction of each pixel point in the first comparison region q respectively. Use the following formula to calculate the comparison value of each pixel point in the first comparison region q respectively: G c = |Gx c | + |Gy c | G c represents the comparison value of pixel point c, Gx c and Gy c respectively represent the detection value in the horizontal direction and the detection value in the vertical direction of pixel point c; Take the pixel points with the comparison value greater than the set comparison value threshold as the elements of EG.

2. The biometric recognition method according to claim 1, characterized in that, Divide the first image into multiple first comparison regions, including: Use a pre - processing algorithm to process the first image to obtain the image PF. Use NL and NW to represent the length and width of PF respectively. Divide PF into N 2 image regions with a length of and a width of , where N represents a preset integer; Perform an enlargement process on the obtained image region to obtain multiple first comparison regions.

3. The biometric recognition method according to claim 2, wherein Perform an enlargement process on the obtained image region to obtain multiple first comparison regions, including: S41, take all the image regions as the elements in the set ARU. S42, randomly extract an image region ru from ARU, perform an enlargement process on ru to obtain a first comparison region, and store the obtained first comparison region in the set CMPU. S43. Determine whether there is still an image area in the ARU. If so, proceed to the second step. If not, use the first comparison area in the CMPU as the determined first comparison area.

4. A biometric recognition method according to claim 3, characterized in that Perform an enlargement process on ru to obtain the first comparison area, including: S51. Use ru as the comparison target and store ru in the set FA. S52. Obtain the set ruA of all image areas adjacent to the comparison target in the ARU. S53. Calculate the competition value of each image area in ruA respectively. S54. Determine whether the maximum value among all the competition values is greater than the set competition value threshold. If so, store the image area with the maximum competition value in ruA into the set FA and proceed to S55. If not, stop the processing and splice all the image areas in FA to obtain the first comparison area. S56. Delete the image areas in FA from the ARU. S57. Use the image area with the maximum competition value in ruA as the updated comparison target and proceed to S52. The calculation formula for the competition value is: cpv r Represents the competition value of the image region r in ruA, egv r Represents the edge judgment value, np v,r And np v,aim Respectively represent the number of pixels with a gray value equal to v in r and the comparison target, Represents the edge judgment weight; The calculation formula for the edge judgment value is: x r and y r are respectively the average values of the abscissas and ordinates of the pixel points belonging to the image edge in the image region r; x aim and y aim are respectively the average values of the abscissas and ordinates of the pixel points belonging to the image edge in the comparison target.

5. A biometric recognition method according to claim 2, characterized in that, Process the first image using a preprocessing algorithm to obtain the image PF, including: Perform noise reduction processing on the first image to obtain an indirect image. Perform enhancement processing on the indirect image to obtain the image PF.

6. The biometric recognition method according to claim 1, wherein For the first comparison area im, the process of obtaining the non-similarity score of im includes: Store the coordinates of all pixel points in im into the set LKim. Obtain the second comparison area formed by the pixel points corresponding to all the coordinates in LKim in the second image PS. Calculate the similarity between im and the second comparison area. Determine whether the similarity is greater than the pre-set similarity threshold. If not, calculate the dissimilarity score usml of im using the following formula im : cmpfir im is the first comparison coefficient for im, CMPU1 is the set of the first comparison regions in P1, cmpfir i represents the first comparison coefficient of the first comparison region i; If so, the non-similarity score of im is 0.

7. A biometric recognition method according to claim 1, characterized in that Calculate the second comparison coefficient of each first comparison area in UP1 based on the first comparison coefficient and UP2 respectively, including: Calculate the second comparison coefficient using the following formula: cmpscd j is the second comparison coefficient for the first comparison region j, cmpfir j is the first comparison coefficient for the first comparison region j, nfir j represents the number of first comparison regions belonging to UP2 among all the first comparison regions adjacent to the first comparison region j, NUP2 represents the total number of first comparison regions in UP2, and λ represents the fourth weighting factor.

8. A biometric recognition system, characterized in that, Include a preprocessing module, an acquisition module, and a comparison module. The preprocessing module is used to calculate the first images containing biometric information pre-entered and stored in the database. Each first image is divided into multiple first comparison areas, and the first comparison coefficient of each first comparison area is calculated respectively. All the first comparison areas and the first comparison coefficients are stored. The acquisition module is used to acquire the second image containing the biometric information of the person whose biometric identification needs to be performed. The comparison module is used to compare the second image with each first image in turn to determine whether there is a first image in the database that meets the identity verification standard. If so, it means that the biometric identification is passed and the comparison ends. Among them, for the first image P1, determine whether P1 meets the identity verification standard through the following method: S30. Use UP1 to represent the set of the first comparison areas corresponding to P1. Obtain the first comparison area ARE1 with the largest first comparison coefficient in UP1 and store ARE1 in the set UP2. S31. Obtain the non-similarity score of ARE1 and delete ARE1 from UP1. S32. Determine whether the sum of the non-similarity scores of the first comparison areas in UP2 is greater than the set score threshold. If so, it means that P1 does not meet the identity verification standard. If not, proceed to S33. S33. Calculate the second comparison coefficient for each first comparison region in UP1 based on the first comparison coefficient and UP2 respectively; S34. Obtain the dissimilarity score of the first comparison region with the largest second comparison coefficient in UP1, delete the first comparison region with the largest second comparison coefficient from UP1, and store the first comparison region with the largest second comparison coefficient in UP2; S35. Determine whether the number of remaining first comparison regions in UP1 is less than 1. If not, go to S32. If so, it means that P1 meets the identity verification standard; The calculation process of the dissimilarity score is as follows: Obtain the second comparison region composed of the pixel points corresponding to the coordinates of all pixel points in the second image that are corresponding to the first comparison region; Calculate the dissimilarity score between the first comparison region and the second comparison region; The calculation formula of the first comparison coefficient is: cmpfir q is the first comparison coefficient for the first comparison region q, neg q is the number of pixel points belonging to the set EG existing in the first comparison region q, np q the number of pixel points in the first comparison region q, length q represents the distance between the coordinates of the center of the first comparison region q and the coordinates of the center of the first image, length max represents the length of the first image, qu is the set of pixel points in the first comparison region q, g z is the image gradient of the pixel point z, g max represents the maximum value of the image gradients of all pixel points in the first comparison regions, d1, d2, and d3 are the first weighting factor, the second weighting factor, and the third weighting factor respectively; The acquisition process of EG is as follows: Calculate the detection value in the horizontal direction and the detection value in the vertical direction for each pixel point in the first comparison region q respectively; Use the following formula to calculate the comparison value for each pixel point in the first comparison region q respectively: G c = |Gx c | + |Gy c | G c represents the comparison value of pixel point c, Gx c and Gy c respectively represent the detection value in the horizontal direction and the detection value in the vertical direction of pixel point c; Take the pixel points with the comparison value greater than the set comparison value threshold as the elements of EG.

Images