A fingerprint comparison method for large-capacity fingerprint database

Abstract

A fingerprint comparison method for a large-capacity fingerprint library, 1. A user registers a fingerprint. 2. A local fingerprint search table is established. 3. The user generates a fingerprint template of the fingerprint to be compared, and then searches for the corresponding feature template in the local template library. The num column in the local fingerprint search table is judged: if the num column is 0, the fingerprint template to be compared is compared with the local template library in a one-to-many comparison. If the comparison is successful, go to step 6; otherwise, the output fingerprint does not exist; if the value in the num column is not 0, go to step 4; 4. According to the number of valid data pieces susNum whose values ​​in the num column in the local fingerprint search table are not 0, the fingerprint template to be compared is compared with the local fingerprint search table according to susNum; 5. The fingerprint template to be compared is compared by using the complement method. 6. The serial number of the fingerprint template to be compared is updated to the local fingerprint search table, and the comparison ends. This method improves the time for searching fingerprints in the fingerprint library and improves the success rate of comparison.

Description

Technical Field

[0001] The invention relates to the field of fingerprint search and comparison, and in particular to a fingerprint comparison method for a large-capacity fingerprint library. Background Art

[0002] When personnel statistics are required in a residential area or factory, each person's fingerprint is generally registered at the property or personnel department and placed in a fingerprint database. Due to the large number of people, the cumulative number of registered users may reach more than 10,000. If a person's fingerprint information needs to be found among them, the conventional practice is to perform a global search each time it is compared with the fingerprint database. This will take a long time, up to more than 2 seconds. Summary of the invention

[0003] In order to compare fingerprints in a large-capacity fingerprint library and improve the comparison efficiency, the present invention adopts the following technical solutions:

[0004] A fingerprint comparison method for a large-capacity fingerprint library comprises the following steps:

[0005] Step S1: The user registers fingerprints and generates fingerprint templates, which are stored in the local template library;

[0006] Step S2: Establish a local fingerprint search table, which includes two columns of data, one column is the sequence number id of successful matching, and the other column is the number of successful matchings num corresponding to each id;

[0007] Step S3: Get the fingerprint template to be compared, and search the local template library for the fingerprint template corresponding to the fingerprint template to be compared: judge the num column in the local fingerprint search table:

[0008] If all values ​​in the num column are 0, the fingerprint template to be compared is compared with the local template library in a one-to-many comparison. If the comparison is successful, proceed to step S6; otherwise, the output fingerprint does not exist;

[0009] If the value in the num column is not 0, go to step S4;

[0010] Step S4: Count the number of valid data entries susNum whose values ​​in the num column in the local fingerprint search table are not 0;

[0011] If susNum is less than the number threshold, the fingerprint template to be compared is compared one-to-one with the local fingerprint search table. If the comparison is successful, the process goes to step S6; otherwise, the fingerprint template to be compared is compared one-to-many with the local template library.

[0012] If susNum is greater than or equal to the number threshold, proceed to step S5;

[0013] Step S5: Compare the fingerprint template to be compared by using the complement method. If the comparison is successful, proceed to step S6; otherwise, the output fingerprint does not exist;

[0014] Step S6: Update the fingerprint template to be compared to the local fingerprint search table, and the comparison is completed.

[0015] Specifically, the fingerprint feature template includes the program version number, the number of feature points, the location information (x, y coordinates), direction, curvature, and feature point type information of each feature point.

[0016] Specifically, in step S5, the process of comparing the fingerprint templates to be compared by the complement method is as follows: the first roughNum templates in the local fingerprint search table are formed into a set preM, and then the fingerprint templates to be compared are compared with the templates in the set preM in a one-to-many manner. If the comparison is successful, proceed to step S6; if the comparison is unsuccessful, the templates remaining in the local template library after removing the set preM are formed into a set extraM, and the fingerprint templates to be compared are compared with the set extraM in a one-to-many manner.

[0017] Specifically, if the value of susNum is greater than one tenth of the total number of templates in the local template library, roughNum is one tenth of the total number of templates in the local template library, otherwise roughNum is equal to susNum.

[0018] Specifically, the one-to-many comparison process is as follows: the set to be compared or the local template library is calculated by the fingerprint similarity method and ranked in descending order of similarity, the first N templates are formed into a set M, and then the fingerprint template to be compared is accurately compared one-to-one with all templates in the set M.

[0019] Specifically, the selection rule of the N value is: the total number of templates in the set to be compared or the local template library is tempSum,

[0020] If tempSum<400, then N=90;

[0021] If 400≤tempSum<800, then N=120;

[0022] If 800≤tempSum<2000, then N=300;

[0023] If 2000≤tempSum<3000, then N=400;

[0024] If 3000≤tempSum<5000, then N=650;

[0025] If 5000≤tempSum<10000, then N=750;

[0026] Otherwise, N=900.

[0027] Specifically, the process of updating the local fingerprint search table is: traverse the local fingerprint search table, find the id in the table that is the same as the fingerprint template to be compared, increase the num value of the corresponding id by 1, and then sort from large to small according to the num value.

[0028] In summary, the present invention and the device have the following advantages: greatly shortening the time of searching for fingerprints in the fingerprint library and improving the success rate of matching. For a local template library of 10,000, the matching time of frequently used users is reduced from 2s to 200ms, which greatly improves the user experience. DETAILED DESCRIPTION

[0029] A fingerprint comparison method for a large-capacity fingerprint library comprises the following steps:

[0030] Step S1: The user registers the fingerprint and generates a fingerprint template. The background record includes the user name, the serial number id mapped one-to-one to the user name, and other information. The fingerprint feature template includes the program version number, the number of feature points, the location information (x, y coordinates), direction, curvature, feature point type, etc. of each feature point. The fingerprint template is stored in the local template library.

[0031] Step S2: Establish a local fingerprint search table, which includes two columns of data, one for the sequence number id of successful matching, and the other for the number of successful matching num corresponding to each id. The local fingerprint search table provides convenience for fingerprint retrieval in a small range. For example, there are many users in a community, and the local fingerprint search table usually collects the matching information of the residents in a fixed unit in the community. When the user of the unit performs fingerprint matching, the local fingerprint search table can be searched first, which can reduce the time for searching and matching in a large range in the local template library and improve the matching efficiency.

[0032] Step S3: The user needs to perform fingerprint comparison to generate a fingerprint template to be compared, and search the local template library for a fingerprint template corresponding to the fingerprint template to be compared: judge the num column in the local fingerprint search table:

[0033] If all values ​​in the num column are 0, the fingerprint template to be compared is compared with the local template library in a one-to-many comparison. If the comparison is successful, proceed to step S6; otherwise, the output fingerprint does not exist;

[0034] If the value in the num column is not 0, go to step S4;

[0035] Step S4: Count the number of valid data entries susNum whose values ​​in the num column in the local fingerprint search table are not 0;

[0036] If susNum is less than the number threshold, which is 90, the serial number id of the fingerprint template to be compared is compared one-to-one with the serial number id in the local fingerprint search table. If the comparison is successful, proceed to step S6; otherwise, the fingerprint template to be compared is compared one-to-many with the local template library.

[0037] If susNum is greater than or equal to the number threshold, which is 90, then go to step S5;

[0038] Step S5: The fingerprint templates to be compared are compared by using the complement method, and the first roughNum templates in the local fingerprint search table form a set preM. If the susNum value is greater than one tenth of the total number of templates in the local template library, then roughNum is one tenth of the total number of templates in the local template library, otherwise roughNum is the susNum value.

[0039] Then the fingerprint template to be compared is compared with the templates in the set preM in a one-to-many comparison. If the comparison is successful, the process proceeds to step S6; if the comparison is unsuccessful, the templates remaining in the local template library after removing the set preM are formed into a set extraM, and the fingerprint template to be compared is compared with the set extraM in a one-to-many comparison. If the comparison is successful, the process proceeds to step S6; otherwise, the output fingerprint does not exist;

[0040] Step S6: Update the fingerprint template to be compared to the local fingerprint search table, obtain the serial number id of the current successful comparison, traverse the local fingerprint search table, find the id in the table that is the same as the fingerprint template to be compared, and increase the num value of the corresponding id by 1, and then sort from large to small according to the num value. It is not necessary to sort the table from beginning to end here, because the table will be sorted every time it is updated, so after updating the num value corresponding to the id, only local sorting is required: search the table upwards to find the serial number id0 with a larger number of times than it: move the data of the id row to the row below the id0 row, and the data after the original id0 is moved down in sequence, and the comparison is completed at the same time.

[0041] The one-to-many comparison process in the above method is as follows: the set to be compared or the local template library is calculated by the fingerprint similarity method, and the reference point pair is found by the similarity of the ridge line where the bifurcation point with higher reliability is located. The reference point pair is selected according to the relationship between the reference point and the surrounding feature points, and the rotation transformation parameters are determined, and then the rotation transformation is performed. After the transformation is completed, all detail points are compared to obtain the similarity score, and the similarity score is ranked in descending order. The first N templates are formed into a set M, and then the fingerprint template to be compared is accurately compared one-to-one with all templates in the set M.

[0042] The selection rule of N value is: the total number of templates in the set to be compared or the local template library is tempSum,

[0043] If tempSum<400, then N=90;

[0044] If 400≤tempSum<800, then N=120;

[0045] If 800≤tempSum<2000, then N=300;

[0046] If 2000≤tempSum<3000, then N=400;

[0047] If 3000≤tempSum<5000, then N=650;

[0048] If 5000≤tempSum<10000, then N=750;

[0049] Otherwise, N=900.

[0050] From the above description, it can be seen that the device has the following advantages: greatly shortening the time of searching for fingerprints in the fingerprint library and improving the success rate of matching. For example, for a local template library of 10,000, the matching time of frequently used users is reduced from 2s to 200ms, which greatly improves the user experience.

[0051] It is understood that the above specific description of the present invention is only used to illustrate the present invention and is not limited to the technical solutions described in the embodiments of the present invention. Those skilled in the art should understand that the present invention can still be modified or replaced by equivalents to achieve the same technical effects; as long as the use requirements are met, they are within the protection scope of the present invention.

Claims

1. A fingerprint comparison method for a large-capacity fingerprint library, characterized in that: The following steps are involved: Step S1: The user registers a fingerprint and generates a fingerprint template, which is stored in a local template library; Step S2: Establish a local fingerprint search table, wherein the local fingerprint search table includes two columns of data, one column is the sequence number id of successful matching, and the other column is the number of successful matchings num corresponding to each id; Step S3: Obtain the fingerprint template to be compared, and search the local template library for a fingerprint template corresponding to the fingerprint template to be compared: determine the num column in the local fingerprint search table: If all values ​​in the num column are 0, a one-to-many comparison is performed between the fingerprint template to be compared and the local template library. If the comparison is successful, the process proceeds to step S6; otherwise, the output fingerprint does not exist. If the value in the num column is not 0, go to step S4; Step S4: Count the number of valid data entries susNum whose values ​​in the num column in the local fingerprint search table are not 0; If susNum is less than the number threshold, the fingerprint template to be compared is compared one-to-one with the local fingerprint search table. If the comparison is successful, the process proceeds to step S6; otherwise, the fingerprint template to be compared is compared one-to-many with the local template library. If susNum is greater than or equal to the number threshold, proceed to step S5; Step S5: Compare the fingerprint template to be compared by the complement method. If the comparison is successful, proceed to step S6; otherwise, the output fingerprint does not exist; Step S6: updating the fingerprint template to be compared to the local fingerprint search table, and the comparison is completed; In step S5, the process of comparing the fingerprint templates to be compared by the complement method is as follows: the first roughNum templates in the local fingerprint search table are formed into a set preM, and then the fingerprint templates to be compared are compared with the templates in the set preM in a one-to-many manner. If the comparison is successful, step S6 is entered; if the comparison is unsuccessful, the templates remaining in the local template library after removing the set preM are formed into a set extraM, and the fingerprint templates to be compared are compared with the set extraM in a one-to-many manner.

2. According to claim 1, a fingerprint comparison method for a large-capacity fingerprint library is characterized in that: The fingerprint feature template includes the program version number, the number of feature points, the position information (x, y) coordinates, direction, curvature, and feature point type information of each feature point.

3. According to claim 1, a fingerprint comparison method for a large-capacity fingerprint library is characterized in that: If the susNum value is greater than one tenth of the total number of templates in the local template library, roughNum is one tenth of the total number of templates in the local template library, otherwise roughNum is equal to susNum.

4. According to claim 1, a fingerprint comparison method for a large-capacity fingerprint library is characterized in that: The one-to-many comparison process is as follows: the set to be compared or the local template library is calculated by the fingerprint similarity method and ranked in descending order of similarity, the first N templates are formed into a set M, and then the fingerprint template to be compared is accurately compared one-to-one with all templates in the set M.

5. According to claim 4, a fingerprint comparison method for a large-capacity fingerprint library is characterized in that: The selection rule of the N value is: the total number of templates in the set to be compared or the local template library is tempSum, If tempSum<400, then N=90; If 400≤tempSum<800, then N=120; If 800≤tempSum<2000, then N=300; If 2000≤tempSum<3000, then N=400; If 3000≤tempSum<5000, then N=650; If 5000≤tempSum<10000, then N=750; Otherwise, N=900.

6. According to claim 1, a fingerprint comparison method for a large-capacity fingerprint library is characterized in that: In step S6, the process of updating the local fingerprint search table is: traversing the local fingerprint search table, finding the id in the table that is the same as the fingerprint template to be compared, and increasing the num value of the corresponding id by 1, and then sorting from large to small according to the num value.