JP2017010419A - Information processing program and information processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information processing program and an information processing device with which it is possible to create verification information that is unsusceptible to the effect of a bio-information acquisition condition even when the size of an image representing bio-information is smaller than a conventional one.SOLUTION: The processor of the information processing device executes the processes of: calculating the geometrical feature values of three feature points extracted from an image; acquiring a plurality of first information corresponding to the position of a first reference point on a line segment linking two points of interest (S37); acquiring, for each of the first information, a sample in which the color information of a second reference point and second information are correlated, for a plurality of second reference points, and acquiring sample information (S38); calculating, for each of the first information, the frequency component of a change in color information to the second information, and acquiring frequency information (S40); and making storage means store verification information in which the sample information and frequency information are correlated with the geometrical feature values (S42).SELECTED DRAWING: Figure 7

Description

  The present invention relates to an information processing program and an information processing apparatus for analyzing images and creating collation information used for collation of biological information.

  In recent years, various fingerprint authentication devices that can be mounted on mobile devices such as smartphones and notebook computers have been studied. For example, Patent Literature 1 discloses a personal identification device that is less susceptible to disturbances such as finger tilt with respect to a fingerprint sensor by using information obtained by frequency spectrum conversion of a fingerprint image as collation information used for collation.

Japanese Patent No. 3057590

  Accompanying the miniaturization of fingerprint sensors mounted on mobile devices, acquired fingerprint images are smaller than before. When a user performs a fingerprint input operation, the finger of the hand holding the mobile device often comes into contact with a fingerprint sensor mounted on the mobile device. In this case, since the user has to move the finger in an unnatural direction, the fingerprint input operation tends to be unstable. That is, it is easy to acquire an image acquired under conditions where the position and angle are different from those at the time of registration. Therefore, there is a demand for a technique for creating collation information that is less affected by the biometric information acquisition condition even when the image size is smaller than that of the conventional art.

  An object of the present invention is to provide an information processing program and an information processing apparatus capable of creating collation information that is not easily affected by the biometric information acquisition condition even when the size of an image representing the biometric information is smaller than the conventional size. It is to be.

  An information processing program according to the first aspect of the present invention is characterized in that a computer having a processor and a storage means analyzes an image acquisition step of acquiring an image, and analyzes the image acquired in the image acquisition step. An extraction step for extracting points; a feature point acquisition step for acquiring three feature points extracted in the extraction step; and a geometry of a triangle represented by the three feature points acquired in the feature point acquisition step A feature amount calculating step for calculating a geometric feature amount; a selection step for selecting two feature points as attention points from the three feature points acquired in the feature point acquisition step; and A first point corresponding to a position of a first reference point, which is a point on a line segment connecting the two selected points of interest and determined based on the geometric feature amount. A reference point acquiring step for acquiring information for each of the plurality of first reference points having different positions, and a plurality of the first information acquired in the reference point acquiring step corresponding to the first information Color information corresponding to a second reference point that is a point on the circumference of a circle having a predetermined radius centered on the first reference point and determined based on the geometric feature amount; and A sample that is information that associates the second information corresponding to the position of the two reference points is acquired for each of the plurality of second reference points that have different positions, and the plurality of acquired samples, A sample information acquisition step for performing processing in which information associated with one information is used as sample information, and based on the sample information acquired in the sample information acquisition step, the plurality of first reference points For each, information is obtained by calculating a frequency component of the color information change with respect to the second information and associating the frequency component calculated for each of the plurality of first reference points with the first information. A frequency information acquisition step for acquiring frequency information, and a storage control for storing the sample information and the frequency information in association with the geometric feature amount in the storage means as verification information used for verification of biological information Is a program for executing steps.

  If collation information is generated based on the information processing program of the 1st mode, collation information including sample information and frequency information is acquired based on the feature point extracted from the image. The sample information and the frequency information are information representing changes in color information of points around a line segment connecting two attention points selected from the extracted feature points. For this reason, even when the number of feature points extracted from the image is small, it is possible to generate information that can be accurately collated. The sample information and the frequency information are information acquired for points determined based on the geometric feature amount. Therefore, if collation information is generated based on this program, information represented by an image (for example, represented by a fingerprint image or a vein image), compared to a case where information is acquired about a point where a geometric feature amount is not considered. It is possible to generate collation information in which the influence of rotation or movement of (biological information) with respect to the reference is reduced.

  The selection step of the information processing program according to the first aspect may select each of three sets of the two attention points corresponding to a combination of all two points of the three feature points. If collation information is generated based on the information processing program in this case, a lot of collation information is generated as compared with a case where a set of two attention points is selected.

  The information processing program according to the first aspect may include a collation step for collating the collation information for testing to be collated with the collation information for reference stored in the storage unit. If biometric information is collated based on the information processing program in this case, sample information and frequency information are acquired based on the feature points extracted from the image, and therefore the number of feature points extracted from the image is small. Can be collated with high accuracy. The sample information and the frequency information are information acquired for points determined based on the geometric feature amount. Therefore, if collation is performed based on this program, the effect of the biological information rotating or moving relative to the reference is reduced compared to the case where information is acquired about the point where the geometric feature amount is not considered. A highly accurate collation result can be obtained.

  The feature point acquisition step of the information processing program according to the first aspect includes the plurality of feature points within the predetermined range determined based on the predetermined radius among the plurality of feature points extracted in the extraction step. In the case of selecting one feature point, a plurality of sets of the three feature points corresponding to all combinations are acquired, and the selection step includes the plurality of sets of the three feature points acquired in the feature point acquisition step. Of these, based on the geometric feature amount calculated in the feature amount calculation step, for the three feature points where the area of the triangle is equal to or greater than a predetermined value, the process of selecting the two attention points, The collation step includes the geometric feature of each of the test triangle represented by the test collation information and the reference triangle represented by the reference collation information. Are extracted in the pair extraction step and the pair extraction step for extracting the pair of the test triangle and the reference triangle whose similarity is equal to or less than a threshold value. The sample information and the frequency information corresponding to the test triangle, and the post-extraction verification step for verifying the sample information and the frequency information corresponding to the reference triangle may be provided.

  When the positions of the three feature points acquired in the feature point acquisition step are close to each other and the area of the triangle represented by the three feature points is less than a predetermined value, the area of the triangle is greater than or equal to the predetermined value Compared to the above, there is little change in the color information indicated by the sample information. Therefore, if biometric information is collated based on the information processing program in this case, if the area of the triangle represented by the three feature points is less than a predetermined value, the collation is performed based on the three feature points. Information is not acquired. Therefore, according to the program of the present invention, it is possible to improve the accuracy of collation using the collation information, compared to the case where the collation information is generated based on three feature points whose triangle area is less than a predetermined value.

An information processing apparatus according to a second aspect of the present invention extracts a feature point by analyzing a storage unit that stores information, an image acquisition unit that acquires an image, and the image acquired by the image acquisition unit. An extraction means; a feature point acquisition means for acquiring three feature points extracted by the extraction means; and a geometric feature amount of a triangle represented by the three feature points acquired by the feature point acquisition means A feature quantity calculating means for calculating the feature point; a selecting means for selecting two feature points as attention points from the three feature points acquired by the feature point acquiring means; and 2 selected by the selecting means. First information corresponding to the position of a first reference point, which is a point on a line segment connecting the two points of interest and determined based on the geometric feature amount, First A reference point acquisition unit that acquires each of the reference points, and a predetermined radius centered on the first reference point corresponding to the first information for each of the plurality of first information acquired by the reference point acquisition unit Color information corresponding to a second reference point that is a point on the circumference of the circle and determined based on the geometric feature, and second information corresponding to a position of the second reference point Is obtained for each of the plurality of second reference points at different positions, and the information obtained by associating the obtained plurality of samples with the first information is used as sample information. Sample information acquisition means for performing processing, and the color for the second information for each of the plurality of first reference points based on the sample information acquired by the sample information acquisition means Frequency information acquisition means for calculating a frequency component of a change in information and acquiring frequency information that is information obtained by associating the frequency component calculated for each of the plurality of first reference points with the first information; And storage control means for storing, in the storage means, information that associates the sample information and the frequency information with the geometric feature value as collation information used for collation of biological information.

  Since the information processing apparatus according to the second aspect acquires sample information and frequency information based on the feature points extracted from the image, the collation information that can be accurately collated even when the number of feature points extracted from the image is small Can be generated. The sample information and the frequency information are information acquired for points determined based on the geometric feature amount. Therefore, the information processing apparatus is based on information represented by an image (for example, biological information represented by a fingerprint image or a vein image), compared to a case where information is acquired about a point where a geometric feature is not considered. On the other hand, it is possible to generate collation information in which the influence of rotation or movement is reduced.

2 is a block diagram of the information processing apparatus 10. FIG. 2 is a functional block diagram of the information processing apparatus 10. FIG. It is a flowchart of collation information processing. It is a flowchart of the image analysis process performed by the collation information processing of FIG. It is the image 41 showing the biometric information for reference acquired from the biometric information acquisition apparatus 8. FIG. It is a figure explaining the process in which the feature points P1 to P11 are extracted from the image 41 of FIG. This is executed in the image analysis process of FIG. It is a flowchart of collation information acquisition processing. It is explanatory drawing of the local coordinate system. It is explanatory drawing of the setting method of a 1st reference point and a 2nd reference point. It is explanatory drawing of the sample 71, the sample data 72, and the sample information 73. FIG. It is explanatory drawing of the collation information 75 for the reference containing the sample information, sample image, frequency information, and frequency image which were acquired for every two attention points. It is explanatory drawing of the process which calculates a frequency component based on the change of the color information with respect to 2nd information represented by sample data. It is explanatory drawing of the collation information registered into DB28. FIG. 3 is a diagram showing a test image 51 and feature points PT1 to PT3 included in the image 51. It is explanatory drawing of the collation information 85 for the test containing the sample information, sample image, frequency information, and frequency image which were acquired for every two attention points about the image 51. FIG. It is a flowchart of the collation process performed by the collation information processing of FIG. It is explanatory drawing of the process in which the collation information for a test and the collation information for a reference are collated. It is explanatory drawing of the process of calculating the similarity of the frequency information for a test, and the frequency information for a reference. It is explanatory drawing of the process of calculating the similarity of the frequency information for a test, and the frequency information for a reference. It is explanatory drawing of the process of calculating the similarity of the frequency information for a test, and the frequency information for a reference. It is explanatory drawing of the process of calculating the similarity of the sample information for a test, and the sample information for a reference. It is explanatory drawing of the process of calculating the similarity of the sample information for a test, and the sample information for a reference. It is explanatory drawing of the process of calculating the similarity of the sample information for a test, and the sample information for a reference.

  An embodiment of the present invention will be described with reference to the drawings. The specific numerical values exemplified in the following embodiments are examples, and the present invention is not limited to these numerical values. In the following description, the image data is simply referred to as “image”.

  The information processing apparatus 10 will be described with reference to FIG. The information processing apparatus 10 is an electronic device having a function of authenticating a user with biometric information. The biological information is selected from various biological information such as face images, fingerprints, veins, and irises. The biometric information of this embodiment is a fingerprint. In the present embodiment, the information processing apparatus 10 is a known smartphone. The information processing apparatus 10 analyzes an image obtained by capturing a fingerprint, generates reference collation information necessary for collation using the fingerprint, and a database (DB) 28 stored in the flash memory 4 of the information processing apparatus 10. The function to memorize is provided. The information processing apparatus 10 analyzes an image obtained by capturing a fingerprint, generates test verification information necessary for verification using the fingerprint, and generates the test verification information and the reference information stored in the DB 28. The function of collating with the collation information is provided.

  As illustrated in FIG. 1, the information processing apparatus 10 includes a CPU 1, a ROM 2, a RAM 3, a flash memory 4, a communication I / F 5, a display unit 6, a touch panel 7, and a biological information acquisition device 8. The CPU 1 is a processor that controls the information processing apparatus 10. The CPU 1 is electrically connected to the ROM 2, RAM 3, flash memory 4, communication I / F 5, display unit 6, touch panel 7, and biological information acquisition device 8. The ROM 2 stores a BIOS, a boot program, and initial setting values. The RAM 3 stores various temporary data. The flash memory 4 stores a program executed by the CPU 1 to control the information processing apparatus 10, an OS (Operating System), and a DB 28. The communication I / F 5 is a controller for executing communication with an external device. The display unit 6 is a liquid crystal display. The touch panel 7 is provided on the surface of the display unit 6. The biometric information acquisition device 8 acquires an image obtained by capturing biometric information. The biometric information acquisition apparatus 8 of this embodiment is a capacitive area type sensor. Specifically, the biometric information acquisition device 8 is a sensor that determines the unevenness of the fingerprint from the amount of charges on the surface electrode in the form of a matrix, and expresses color information as a gradation value of 256 gradations for each pixel. The color information is information representing a color. The resolution is, for example, 508 dpi (dots per inch).

  An overview of functions of the information processing apparatus 10 will be described with reference to FIG. The information processing apparatus 10 includes a biological information acquisition device 8, an image acquisition unit 21, a feature point extraction unit 22, a geometric feature amount calculation unit 23, a sample information acquisition unit 24, a frequency information acquisition unit 25, a collation unit 27, and a registration unit. 26 and DB 28, and processing corresponding to each functional block is executed by the CPU 1 (see FIG. 1).

  The biometric information acquisition device 8 outputs an image to the image acquisition unit 21. The image acquisition unit 21 acquires the image output from the biological information acquisition device 8. The feature point extraction unit 22 extracts the two-dimensional coordinates of the feature points in the image coordinate system based on the image acquired from the image acquisition unit 21. The two-dimensional coordinates in the image coordinate system of the present embodiment are coordinates set in units of pixels based on the positions of pixels in the image. The two-dimensional coordinates of the image coordinate system will be described later. The feature points are, for example, points extracted according to the minutiae extraction method (see, for example, Japanese Patent No. 1289457) and SIFT (Scale-Invariant Feature Transform) (see, for example, US Pat. No. 6,711,293). ) At least one of the points extracted according to. The feature points of this embodiment are extracted according to the minutiae extraction method described in Japanese Patent No. 1289457.

  The geometric feature quantity calculation unit 23 acquires the coordinates of the feature points within a predetermined range among the plurality of feature points extracted by the feature point extraction unit 22. The geometric feature amount calculation unit 23 selects three feature points from a plurality of feature points whose coordinates are acquired, and calculates a geometric feature amount of a triangle represented by the selected three feature points. To do. The geometric feature amount is, for example, the coordinates of three feature points, the slope (vector) of a line segment represented by two attention points selected from the three feature points, the length of the sides of the triangle, the triangle Are appropriately selected from the interior angles and the triangular areas. The coordinates of the three feature points are represented by absolute coordinates or relative coordinates in the image coordinate system. The geometric feature amount calculation unit 23 of the present embodiment sets a local coordinate system for the selected three feature points in addition to the coordinates of the image coordinate system, and the coordinates of the three feature points also in the coordinates of the local coordinate system. Represents. The local coordinate system will be described later. The geometric feature quantity calculation unit 23 of the present embodiment acquires a plurality of sets of three feature points corresponding to all combinations when selecting three feature points from a plurality of feature points within a predetermined range. , Calculate geometric features for each.

  The sample information acquisition unit 24 acquires sample information based on the geometric feature amounts of the three feature points acquired by the geometric feature amount calculation unit 23 and predetermined acquisition conditions. The sample information is information in which color information is associated with information (first information and second information) corresponding to a position indicating the acquisition condition of the color information. The sample information is acquired by the following procedure, for example. Of the three feature points, two feature points are respectively selected as points of interest. The sample information acquisition unit 24 of the present embodiment is a triangle represented by three feature points based on a geometric feature amount among a plurality of sets of three feature points acquired by the geometric feature amount calculation unit 23. A process of selecting two attention points is executed for three feature points whose area is equal to or greater than a predetermined value. The sample information acquisition unit 24 determines a point on the line segment connecting the two selected points of interest and determined based on the geometric feature amount as the first reference point. First information is acquired for each of a plurality of first reference points whose positions are different from each other. The first information is information corresponding to the position of the first reference point. The first information may be information that defines the position of the first reference point with respect to the standard. For example, absolute coordinates (for example, coordinates in the image coordinate system), relative coordinates (for example, coordinates in the local coordinate system), and standard It may be an angle with respect to. When the acquisition order of the first reference points is determined with respect to the standard, the first information may be the acquisition order of the first reference points. A plurality of samples are acquired for each of the acquired plurality of pieces of first information. The sample is information in which the color information corresponding to the second reference point is associated with the second information. The second reference point is a point on the circumference of a circle having a predetermined radius centered on the first reference point corresponding to the first information, and is determined based on the geometric feature amount. The second information is information corresponding to the position of the second reference point. The second information may be information that defines the position of the second reference point with respect to the standard. For example, absolute coordinates (for example, coordinates in the image coordinate system), relative coordinates (for example, coordinates in the local coordinate system), and standards It may be an angle with respect to. When the acquisition order of the second reference points is determined with respect to the standard, the second information may be the acquisition order of the second reference points. A plurality of samples are samples acquired about each of a plurality of 2nd reference points from which a position mutually differs. Information obtained by associating the plurality of acquired samples with the first information is sample information.

  The frequency information acquisition unit 25 acquires frequency information based on the sample information acquired from the sample information acquisition unit 24. For the frequency information, for each of the plurality of first reference points, the frequency component of the color information change with respect to the second information is calculated, and the frequency components calculated for each of the plurality of first reference points are associated with the first information. Information. The frequency component of the present embodiment is a one-dimensional group delay spectrum calculated in accordance with a known method (see Japanese Patent Nos. 3057590 and 3799057).

  The registration unit 26 stores reference verification information used for verification of biometric information in the DB 28. The collation information is information in which sample information and frequency information are associated with geometric feature amounts. In the present embodiment, collation information is stored for each of the three feature points of all combinations selected from effective feature points. The collation unit 27 collates the test collation information with the reference collation information stored in the DB 28. More specifically, the collation unit 27 calculates the geometric feature amounts of the test triangle represented by the test collation information and the reference triangle represented by the reference collation information as triangles. Compared in descending order of area, a pair of a test triangle and a reference triangle having a similarity equal to or less than a threshold is extracted. The collating unit 27 collates the sample information and frequency information corresponding to the extracted test triangle with the sample information and frequency information corresponding to the reference triangle.

1. Processing at the time of registration With reference to FIG. 3, a case where collation information is registered will be described as an example for collation information processing executed by the information processing apparatus 10. The verification information processing is started when the user inputs a start instruction. The start instruction includes an instruction regarding whether the collation information acquired from the image is registered in the DB 28 as reference collation information or whether the collation information is collated with the reference collation information registered in the DB 28. When the CPU 1 of the information processing apparatus 10 detects the input of the collation information processing start instruction, the CPU 1 reads out the information processing program for executing the collation information processing stored in the flash memory 4 into the RAM 3, and the instruction included in the information processing program The processing of each step described below is executed according to the above. In the present embodiment, feedback processing that prompts re-input is executed until biometric information that satisfies the requirement for extracting feature points (for example, image clarity) is acquired. The biometric information acquired by the collation information processing satisfies the requirement for extracting the collation information from the biometric information using an algorithm. Information and data acquired or generated in the course of processing are stored in the RAM 3 as appropriate. Various setting values necessary for processing are stored in the flash memory 4 in advance. Hereinafter, the step is abbreviated as “S”.

  As shown in FIG. 3, the CPU 1 executes image analysis processing (S1). The image analysis process will be described with reference to FIG. When the biometric information acquisition device 8 detects a finger contact, the biometric information acquisition device 8 outputs to the CPU 1 a signal that can specify an image obtained by capturing a fingerprint. The CPU 1 receives a signal output from the biological information acquisition device 8. As shown in FIG. 4, the CPU 1 acquires an image based on the received signal (S11). In S11, for example, an image 41 shown in FIG. 5 is acquired. In the image 41, two-dimensional coordinates of an image coordinate system indicated by Xp and Yp in FIG. 5 are set. The position of the upper left pixel of the image 41 is the origin of the two-dimensional coordinate 45 (see FIG. 6) of the image coordinate system. The position of a pixel that is separated by x pixels in the Xp plus direction from the origin of the two-dimensional coordinate 45 and separated by y pixels in the Yp plus direction from the origin is denoted as coordinates (x, y). The image 41 is a square image with a 192 pixel in the Xp direction (left and right direction) and 192 pixels in the Yp direction (up and down direction), which represents a 1 cm square square imaging range. In the two-dimensional coordinates 45 of the image coordinate system, each of the points P1 to P11 is within a range 43 representing the size of the image 41. Although not shown, well-known preprocessing is appropriately performed on the image acquired in S11.

  The CPU 1 extracts feature points from the image acquired in S11 (S12). The CPU 1 of the present embodiment represents a point representing a dead end point of a ridge that is a series of convex portions of a fingerprint and a branching point divided into two according to a known method (for example, see Japanese Patent No. 1289457). Each point is extracted as a feature point. As shown in FIG. 6, from the image 41, points representing end points and points representing branch points surrounded by a circle 42 are extracted as feature points. Eleven feature points extracted from FIG. 41 are designated as points P1 to P11.

  The CPU 1 extracts a valid feature point from the feature points extracted in S12, and executes a process of obtaining the number J of valid feature points (J is an integer equal to or greater than 0) (S13 to S16). The CPU 1 according to the present embodiment sets a feature point within the range 44 as an effective feature point, and sets a feature point outside the range 44 as an invalid feature point. The range 44 is a rectangular range whose four sides are 15 pixels inside the rectangle indicating the range 43 representing the size of the image 41. The reason why the feature point in the range 44 is an effective feature point is to acquire a feature point that can set a second reference point to be described later in the range 43. Information for specifying the range 44 is determined in advance based on a second reference point acquisition condition described later, and is stored in the flash memory 4.

  Specifically, the CPU 1 acquires one coordinate of the feature point that has not been subjected to the process of S13 from the feature points extracted in S12 (S13). The CPU 1 determines whether or not the coordinates acquired in S13 are within the range 44 (S14). The CPU 1 sets points inside the boundary line of the range 44 and points on the boundary line of the range 44 within the range 44. When the coordinates are within the range 44 (S14: YES), the CPU 1 sets the feature point indicated by the coordinates acquired in S13 as an effective feature point, and increments the number J by 1 (S15). The initial value of the number J is 0. The points P1 to P11 in the specific example are all within the range 44 (S14: YES). Therefore, the number J in the specific example finally becomes 11. When the coordinates are outside the range 44 (S14: NO), or after S15, the CPU 1 determines whether or not the process of S13 has been executed for all the feature points extracted in S12 (S16). If there is a feature point whose coordinates have not been acquired in the process of S13 (S16: NO), the CPU 1 returns the process to S13. When the process of S13 is executed for all the feature points extracted in S12 (S16: YES), the CPU 1 determines whether or not the number J is larger than 2 (S17). When the number J is 2 or less (S17: NO), the CPU 1 ends the image analysis process and returns the process to the collation information process of FIG.

The number J in the specific example is 11, which is larger than 2 (S17: YES). In this case, the CPU 1 calculates all combinations when selecting three feature points from among the effective feature points of several J (S18). In the specific example, the number J is 11, and the CPU 1 calculates 165 combinations of ₁₁ C ₃ . The CPU 1 selects one combination for which S19 has not been executed from the combinations calculated in S18, and acquires the coordinates of the three feature points indicated by the selected combination (S19). The case where the CPU 1 acquires two-dimensional coordinates of the image coordinate system for each of the points P1, P4, and P8 in FIG. 6 as three feature points will be described as a specific example. The CPU 1 calculates the area of a triangle connecting the three feature points based on the coordinates acquired in S19 (S20). The CPU 1 determines whether or not the area calculated in S20 is larger than a threshold value (S21). S20 and S21 are processes for extracting three feature points to be subjected to a collation information acquisition process described later. The three feature points acquired in S19 may be located relatively close to each other, for example, points P1, P2, and P3. The points P1, P2, and P4 may be on substantially the same straight line. In these cases, when collation information is acquired according to the procedure described later, there is a possibility that the number of samples decreases or a plurality of similar samples are acquired, and collation information suitable for collation cannot be obtained. For this reason, CPU1 of this embodiment does not acquire collation information based on the coordinate of these three feature points, when the area of a triangle becomes below a threshold value. When the area is equal to or smaller than the threshold (S21: NO), the CPU 1 discards the combination of the three feature points whose coordinates have been acquired in S19 (S24), and executes the process of S25 described later.

  In the specific example, it is determined that the area is larger than the threshold (S21: YES), and the CPU 1 executes a collation information acquisition process (S22). The collation information acquisition process is a process for obtaining collation information used for collation of biometric information based on the coordinates of the three feature points acquired in S19. As shown in FIG. 7, in the collation information acquisition process, first, the CPU 1 calculates distances d between two points included in the three feature points based on the coordinates acquired in S19 of FIG. 4 (S32). The distance d is the length of each side of the triangle represented by the three feature points. In the specific example, the distance between the point P1 and the point P4, the distance between the point P4 and the point P8, and the distance between the point P8 and the point P1 are calculated.

  CPU1 sets a local coordinate system based on the calculation result of S32 (S33). The local coordinate system is a relative coordinate set for each triangle based on the coordinates of the three feature points acquired in S19. Specifically, the CPU 1 sets the Xk axis of the local coordinate system on a line segment connecting two points having the longest distance among the three distances d calculated in S32. The CPU 1 sets the side with the remaining one point as the positive direction of the Yk axis, and sets the point on the left side of the two points on the Xk axis as the origin. In the specific example, a local coordinate system 60 is set for the points P1, P4, and P8 as shown in FIG. Of the sides LA, LB, and LC included in the triangle TRR formed by the points P1, P4, and P8, the side LA is the longest. That is, the distance between the point P4 and the point P8 is the longest. Therefore, the Xk axis is set on the side LA, the side where the point P1 is located with respect to the side LA is the positive direction of the Yk axis, and the point P8 on the left side of the points P4 and P8 is the origin.

  The CPU 1 acquires two feature points of the combination not acquired in S34 as attention points from the three feature points (S34). The CPU 1 first acquires two points (two points on the Xk axis) forming the base in the local coordinate system, and thereafter acquires a combination of the two points clockwise. In the specific example, the CPU 1 acquires two attention points in the order of the combination of the point P4 and the point P8, the combination of the point P8 and the point P1, and the combination of the point P1 and the point P4. The CPU 1 calculates a direction vector V from the start point to the end point with one of the two attention points acquired in S34 as the start point and the other as the end point (S35). The CPU 1 sets the point on the upstream side when the triangle side formed by the three feature points is traced clockwise as the start point and the point on the downstream side as the end point. The CPU 1 calculates the direction vector V using the two-dimensional coordinates of the image coordinate system. In the combination of P4 and point P8, the CPU 1 according to the present embodiment uses the vector VA from point P4 to point P8, and the combination of point P8 and point P1 to vector VB from point P8 to point P1, point P1, and point P4. In the combination, each vector VC from the point P1 to the point P4 is calculated (S35).

  CPU1 calculates value L (S36). The value L is a positive integer obtained by rounding down the decimal point of the distance d between the two attention points acquired in S34. The CPU 1 determines the first reference point based on the geometric feature amount of the triangle, and calculates the coordinates of the set first reference point (S37). The coordinates of the first reference point are acquired as two-dimensional coordinates in the image coordinate system. The CPU 1 of the present embodiment sets (L + 1) first reference points on the line segment connecting the two attention points acquired in S34. The (L + 1) first reference points are both ends of a line segment connecting two attention points (that is, two attention points), and points at which the two attention points are equally divided by an interval U. The interval U is a value obtained by dividing the distance d between two attention points by a number L. In this embodiment, the interval U is approximately 1. The CPU 1 sequentially sets the i-th first reference point at equal intervals U from the start point to the end point of the line connecting the two attention points using the direction vector calculated in S35 and the coordinates of the start point. To do. i is a natural number from 1 to (L + 1).

  The CPU 1 acquires sample information (S38). The CPU 1 according to the present embodiment acquires sample information for each of two attention points for each of the three feature points of all combinations calculated in S18 of FIG. A method for acquiring sample information will be described using a specific example. The CPU I1 sets a plurality of second reference points based on the coordinates of the first reference points acquired in S37, and calculates the coordinates of the set plurality of second reference points. In the present embodiment, 128 second reference points are set at equal intervals on the circumference of a circle having a radius of 15 pixels centered on the first reference point. The radius R of the circle and the number N of the second reference points may be appropriately set in consideration of the image size, resolution, authentication accuracy, processing speed, and the like. The number N of the second reference points is preferably set to a power of 2 from the viewpoint of acquiring frequency information described later.

  The CPU 1 of the present embodiment calculates the coordinates of the j-th second reference point corresponding to the i-th first reference point as follows. j is a natural number from 1 to 128 (N). The point where j is 1 is called the starting point SP. As shown in FIG. 9, when the two points of interest are points P4 and P8, the first reference point P4 is on a circle C1 having a radius R centered on the point P4, and the first reference point P4. A point Q11 in the vector VA direction is set as the starting point SP. For the first reference point P8, a point Q12 on the circle C2 having a radius R centered on the point P8 and in the vector VA direction with respect to the first reference point P8 is set as the start point SP. When the two attention points are points P8 and P1, the point Q21 on the circle C2 with respect to the first reference point P8 and in the vector VB direction with respect to the first reference point P8 is the start point SP. Is set. For the first reference point P1, a point Q22 on the circle C3 having a radius R centered on the point P1 and in the vector VB direction with respect to the first reference point P1 is set as the start point SP. When the two points of interest are the points P1 and P4, the point Q31 on the circle C3 with respect to the first reference point P1 and in the vector VC direction with respect to the first reference point P1 is set as the start point SP. Is done. For the first reference point P4, a point Q32 on the circle C1 and in the vector VC direction with respect to the first reference point P4 is set as the start point SP. The second reference points where j is 2 or later are set at every sampling interval counterclockwise from the starting point SP. When the sampling interval of the present embodiment is represented by an angle with respect to a circle, 360 degrees (2π radians) is divided by 128.

  CPU1 acquires a sample about each of a plurality of 2nd reference points. A sample acquisition method will be described with reference to FIG. The CPU 1 of the present embodiment acquires the gradation value of the pixel represented by the second reference point as color information. When the coordinates of the second reference point are in sub-pixel units (when the absolute coordinates are not an integer), the CPU 1 uses the tone values of the four pixels in the vicinity of the second reference point by the well-known bilinear interpolation method to calculate the second reference point. Reference point color information is calculated. The CPU 1 uses, as a sample 71, information obtained by associating the color information acquired for one second reference point with the second information corresponding to the second reference point. The second information of the present embodiment is j representing the acquisition order corresponding to the position of the second reference point with respect to the start point SP. That is, the second information of the present embodiment is represented by a natural number from 1 to 128.

  As schematically illustrated in FIG. 10, the CPU 1 sets a plurality of samples 71 acquired for one first reference point as sample data 72 in association with the first information. The CPU 1 acquires the sample information 73 by performing the above processing for each of (L + 1) first reference points set for one set of two attention points. That is, the sample information 73 includes (L + 1) sample data 72 calculated for each of (L + 1) first reference points. Each sample data 72 includes N samples 71 acquired for each of the N second reference points. Sample information 73 is acquired for every two points of interest. In the specific example, as shown in FIG. 11, in the specific example, sample information SDR1 is acquired for points P4 and P8, sample information SDR2 is acquired for points P8 and P1, and sample information SDR3 is acquired for points P1 and P4. The

  The CPU 1 generates a sample image based on the acquired sample information (S39). In the specific example, sample images 76 to 78 shown in FIG. 11 are generated for each of the three sets of two attention points. The longitudinal directions of the sample images 76 to 78 respectively correspond to the first information, and the lateral direction corresponds to the second information. The color of each pixel in the sample images 76 to 78 represents the color information of the pixel corresponding to the combination of the first information and the second information.

  The CPU 1 acquires frequency information based on the sample information acquired in S38 (S40). The CPU 1 according to the present embodiment calculates, as frequency information, a one-dimensional group delay spectrum of changes in color information with respect to the second information for each first information according to the above-described known method. The group delay spectrum (GDS) is defined as the frequency derivative of the phase spectrum in the power transfer function. As illustrated in FIG. 12, the GDS 83 calculated based on the sample information 82 separates and emphasizes individual peaks of the frequency spectrum of the sample information 82. The number of elements in the GDS array is obtained by subtracting 1 from the number of elements in the phase spectrum. Specifically, the CPU 1 calculates the GDS by taking the phase derivative of the power spectrum obtained by performing the fast Fourier transform on the weighted LPC coefficient, and uses the calculated GDS as the frequency component. In the specific example, as shown in FIG. 11, frequency information FDR1 is acquired for points P4 and P8, frequency information FDR2 is acquired for points P8 and P1, and frequency information FDR3 is acquired for points P1 and P4.

  CPU1 produces | generates a frequency image based on the acquired frequency information (S41). In the specific example, frequency images 79 to 81 shown in FIG. 11 are generated. The CPU 1 makes a pixel having a small GDS intensity light (decrease the gradation value). In a specific example, the short direction of the frequency images 79 to 81 corresponds to the number of elements, and the long direction corresponds to the second information. The color of each pixel in the frequency images 79 to 81 represents the GDS intensity of the pixel corresponding to the combination of the number of elements and the second information. The CPU 1 darkens the pixels having a relatively large GDS intensity (increases the gradation value). The CPU 1 makes a pixel having a relatively small GDS intensity light (decrease the gradation value).

  The CPU 1 stores information in which the sample information and the frequency information are associated with the geometric feature amount in the RAM 3 as collation information used for collation of the biological information (S42). As shown in FIG. 11, in the specific example, collation information 75 in which sample information SDR1 to SDR3 and frequency information FDR1 to FDR3 are associated with the geometric feature amount GDR is stored in the RAM 3. The geometric feature amount GDR includes the area calculated in S20, the distance calculated in S32, the coordinates of the local coordinate system set in S33, the direction vector V calculated in S35, and the interior angle of the triangle. The collation information 75 of this embodiment includes the sample image generated in S39 of FIG. 7 and the frequency image generated in S41. The CPU 1 ends the collation information acquisition process, and returns the process to the image analysis process shown in FIG.

  As shown in FIG. 3, after S22, the CPU 1 determines whether or not three feature points of all combinations calculated in S18 have been acquired in the process of S19 (S25). When there is a combination that has not been acquired (S25: NO), the CPU 1 returns the process to S19. When three feature points of all combinations are acquired in the process of S19 (S25: YES), the CPU 1 ends the image analysis process and returns the process to the collation information process of FIG.

  CPU1 judges whether collation information was acquired by S1 (S2). When collation information is not acquired (S2: NO), CPU1 notifies an error (S6). For example, the CPU 1 displays an error message on the display unit 6. When the collation information is acquired (S2: YES), it is determined whether or not the collation information acquired in S1 is registered in the DB 28 (see FIG. 2) as reference collation information (S3). Information on whether or not to register is included in the start instruction, for example. In the specific example, it is determined to be registered (S3: YES), and the CPU 1 stores the collation information acquired in S1 in the DB 28 of the flash memory 4. In the present embodiment, collation information is acquired for combinations that satisfy the requirements of S21 among the three feature points of all combinations calculated in S18 of FIG. In a specific example, as shown in FIG. 13, a plurality of pieces of collation information including collation information 75 in the case where the three feature points are points P1, P4, and P8 are registered in the DB. When not registering (S3: NO), CPU1 performs the collation process which makes the collation information acquired by S1 the collation information for the test used as the object of collation (S5). Next to any of S4, S5, and S6, the CPU 1 ends the collation information processing.

2. Processing at the time of collation For example, the collation information extracted from the image 41 in FIG. 5 is used as the reference collation information, and the image 51 in FIG. 14 is acquired as the test image to be collated. Will be described. In the collation information processing at the time of collation, S1 is executed similarly to the collation information processing at the time of registration. For example, if the points PT1, PT2, and PT3 in FIG. 14 are acquired as three feature points in S19 in FIG. 4, SDT3 is acquired from the sample information SDT1 in FIG. 15 in S38 in FIG. To 88 are generated. In S40, frequency information FDT1 to FDT3 are acquired, and in S41, frequency images 89 to 91 are generated. In S42, collation information 85 is stored.

  In S2, it is determined that the collation information has been acquired (S2: YES), and it is determined not to register based on the start instruction (S3: NO). CPU1 performs a collation process (S5). As shown in FIG. 14, in the collation process, the CPU 1 sorts collation information corresponding to each of a plurality of sets of three feature points based on the area S of the triangles included in the collation information. Are rearranged (S51). The CPU 1 executes the process of S51 for each of the reference verification information stored in the DB 28 and the test verification information acquired in S1 of FIG. When the reference collation information stored in the DB 28 is sorted in advance, the CPU 1 sorts only the test collation information acquired in S1 of FIG. As schematically shown in FIG. 17, by the processing in S <b> 51, for each of the verification information for testing and the verification information for reference, a plurality of sets of three feature points included in the verification information are converted into three feature points. Are sorted in descending order of the area S of the triangles.

  The CPU 1 selects one set of three feature points in descending order of the area S from a plurality of sets of three feature points included in the test verification information (S52). The CPU 1 selects a set of three feature points in descending order of the area S from a plurality of sets of three feature points included in the reference matching information (S53). The CPU 1 includes a geometric feature amount associated with the three test feature points acquired in S52, and a geometric feature amount associated with the three reference feature points acquired in S53. Is calculated (S54). The process of S54 includes a triangle formed by the three feature points for testing acquired in S52 (hereinafter referred to as “test triangle”) and a triangle formed by the three feature points for reference acquired in S53 (hereinafter referred to as “triangle for testing”). , “Reference triangle”) and the like. In the present embodiment, the coordinates of the local coordinate system and the interior angle of the triangle are used as the geometric feature amount in S54.

  The CPU 1 determines whether or not the geometric feature amount difference calculated in S54 is smaller than a threshold value (S55). When the difference between the coordinates of the local coordinate system for each of the three feature points is greater than or equal to the threshold value, and / or when the difference between the interior angles of the triangles is greater than or equal to the threshold value (S55: NO), the CPU 1 performs geometric Processing is performed when the test triangle and the reference triangle are not similar based on the target feature amount. Specifically, for the test triangle, the CPU 1 determines whether or not the difference in length between the first long side and the second long side among the three sides is larger than the threshold (S63). The process of S63 is performed when the triangle represented by the three test feature points is a triangle close to an isosceles triangle in which the difference in length between the first long side and the second long side is equal to or less than a threshold value. This is a process that takes into account. When the difference in length between the first long side LK1 and the second longest side LK2 is equal to or less than the threshold value as in the triangle TRK formed by the points PK1, PK2, and PK3 in FIG. 17 (S63: NO) The CPU 1 resets the coordinates of the local coordinate system for the three test feature points (S64). Specifically, the CPU 1 sets the local coordinate system 61 in which the second long side LK2 is arranged on the Xk axis, and the side where the point PK2 is located is the Yk axis plus side (S64). The CPU 1 returns the process to S54.

  When the difference in length between the first long side and the second long side is larger than the threshold value (S63: YES), the CPU 1 reads the last three feature points for reference selected in S53 in the reading order. It is determined whether the three feature points are (S65). The CPU 1 of this embodiment reads three feature points in descending order of the triangular area S from the reference information for reference. When the reading order is not the last (S65: NO), the CPU 1 returns the process to S53. When the reading order is last (S65: YES), the CPU 1 discards the three feature points for testing selected in S52 (S66). The CPU 1 determines whether or not the three feature points for testing selected in S52 are the last feature points in the reading order (S67). The CPU 1 of the present embodiment reads three feature points in descending order of the triangular area S from the test verification information. When the reading order is not the last (S67: NO), the CPU 1 returns the process to S52. When the reading order is last (S67: YES), the CPU 1 determines that the authentication has failed (S68). CPU1 complete | finishes collation process and returns a process to the collation information processing of FIG. CPU1 complete | finishes collation information processing after S5.

  In S55, when the difference in coordinates of the local coordinate system for each of the three feature points is smaller than the threshold and the difference in the interior angle of the triangle is smaller than the threshold (S55: YES), the CPU 1 calculates the similarity of the frequency information. (S58). Specifically, the CPU 1 determines the similarity between the frequency information acquired for the three test feature points selected in S52 and the frequency information acquired for the three reference feature points selected in S53. Is calculated. As a method for calculating the similarity, a known method may be used as appropriate. The CPU 1 according to the present embodiment calculates a distance value as a similarity using a known DP matching. The CPU 1 of the present embodiment calculates the distance value of the frequency information for each of three sets of two attention points selected from the three feature points.

Specifically, as shown in FIG. 17, when the reference triangle is the triangle TRR and the test triangle is the triangle TRT, the reference frequency information is the frequency information FDR1 to FDR3 in FIG. The frequency information is frequency information FDT1 to FDT3 in FIG. In this case, as schematically shown in FIG. 18, the CPU 1 uses the frequency image 79 and the frequency image 89 for DP matching to obtain a distance value DF1. In the present embodiment, the matching cost is obtained based on 14 components corresponding to 1 to 14 selected with priority on the low-order components in order to eliminate noise. In FIG. 18, the frequency image 89 representing the test frequency information FDT1 is shown on the upper side. In FIG. 18, a frequency image 79 representing the reference frequency information FDR1 is shown on the left side in a direction corresponding to the frequency image 89. The direction corresponding to the frequency image 89 is that the frequency image 79 is inverted about the lower end of the frequency image, and then the inverted frequency image 79 is rotated 90 degrees clockwise from the upper left end of the inverted frequency image 79. Can be obtained. The CPU 1 starts a path search from the upper left corresponding point, searches for a corresponding point with the lowest matching cost among adjacent corresponding points, and searches for a path reaching the lower right corresponding point. In FIG. 18, the path is indicated by a thick line 95 extending substantially linearly from the upper left to the lower right. The same applies to the diagrams schematically showing the following DP matching. As schematically shown in FIG. 19, the CPU 1 uses the frequency image 80 and the frequency image 90 for DP matching to obtain a distance value DF2. As schematically shown in FIG. 20, the CPU 1 uses the frequency image 81 and the frequency image 91 for DP matching to obtain a distance value DF3. The CPU 1 calculates a standardized distance value DFD based on the calculated three distance values.
DFD = (DF−DFH) / DFS (1)
In Expression (1), DF is an actual distance value (any one of DF1 to DF3) obtained by DP matching of frequency information. DFH is an average value of distance values obtained by DP matching of frequency information. DSS is a standard deviation of distance values obtained by DP matching of frequency information. The DFH and DFS of this embodiment are values set in advance based on the DP matching results of a plurality of frequency information. Based on the equation (1), the standardized distance values DFD1 to DFD3 are calculated from the distance values DF1 to DF3. The distance value DFD has a smaller similarity as the value is larger than when the value is smaller.

The CPU 1 calculates the similarity of the sample information (S59). Specifically, the CPU 1 determines the similarity between the sample information acquired for the three feature points for testing selected in S52 and the sample information acquired for the three feature points for reference selected in S53. Is calculated. As a method for calculating the degree of similarity, a known method may be selected as appropriate. In the present embodiment, as in the case of frequency information, a known DP matching is used, and a distance value is calculated as a similarity. The CPU 1 calculates the distance value of the sample information for each of the three sets of two attention points selected from the three feature points. In the specific example, the reference sample information is sample information SDR1 to SDR3 in FIG. 11, and the test sample information is sample information SDT1 to SDT3 in FIG. In this case, as schematically shown in FIG. 21, the CPU 1 uses the sample image 76 and the sample image 86 for DP matching to obtain a distance value DS1. As schematically shown in FIG. 22, the CPU 1 uses the sample image 77 and the sample image 87 for DP matching to obtain a distance value DS2. As schematically shown in FIG. 23, the CPU 1 uses the sample image 78 and the sample image 88 for DP matching to obtain a distance value DS3. The CPU 1 calculates a standardized distance value DSD according to the following equation (2) for each of the calculated three distance values.
DSD = (DS−DSH) / DSS (2)
In Equation (2), DS is an actual distance value (any one of DS1 to DS3) obtained by DP matching of sample information. DSH is an average value of distance values obtained by DP matching of sample information. DSS is a standard deviation of distance values obtained by DP matching. DSH and DSS of the present embodiment are values set in advance based on the DP matching results of a plurality of sample information. Based on the equation (2), standardized distance values DSD1 to DSD3 are calculated from the distance values DS1 to DS3. The distance value DSD has a smaller similarity as the value is larger than when the value is smaller.

The CPU 1 calculates a similarity D used for collation based on the similarity of the frequency information calculated in S58 and the similarity of the sample information calculated in S59 (S60). Specifically, the CPU 1 adds values obtained by multiplying each of the similarity of the frequency information calculated in S58 and the similarity of the sample information calculated in S59 by a coefficient. The coefficient is set in advance in consideration of the authentication result. The CPU 1 of the present embodiment calculates the similarity D according to the equation (3). However, B is a number from 0 to 1. The similarity D is calculated for every two points of interest. For example, for the points P4 and P8, the similarity D1 is calculated based on B × DSD1 + (1−B) × DFD1.
D = B × DSD + (1−B) × DFD Formula (3)

  The CPU 1 determines whether or not the similarity DR is smaller than a threshold value (S61). In the present embodiment, the representative value of the three similarities D calculated for each two attention points in S60 is set as the similarity DR used for the comparison in S61. The similarity DR is, for example, one of an average value, a minimum value, an intermediate value, and a maximum value of three similarities. The threshold value in S61 is appropriately set according to the representative value. When the similarity DR is smaller than the threshold (S61: YES), the CPU 1 determines that the authentication is successful (S62). If the similarity is equal to or greater than the threshold (S61: NO), the CPU 1 determines that the authentication has failed (S68). After the process of S62 or S68, the CPU 1 ends the collation process and returns the process to the collation information process of FIG. CPU1 complete | finishes collation information processing above. Although not shown, when an authentication experiment using the verification information of the present embodiment and an authentication experiment performed by a conventional method were performed, the verification accuracy equal to or higher than the conventional one was confirmed by the verification information processing of the present embodiment.

  In the information processing apparatus 10, the CPU 1 corresponds to the processor of the present invention. The RAM 3 and the flash memory 4 correspond to storage means of the present invention. The process of S11 in FIG. 4 corresponds to the image acquisition step of the present invention. The CPU 1 functioning as the image acquisition unit 21 in S11 corresponds to the image acquisition unit of the present invention. The process of S12 corresponds to the extraction step of the present invention. The CPU 1 functioning as the feature point extraction unit 22 in S12 corresponds to the extraction unit of the present invention. The process of S19 corresponds to a feature point acquisition step of the present invention. CPU1 which performs the process of S19 functions as a feature point acquisition means of this invention. The processes of S20, S32 and S35 of FIG. 7 correspond to the feature amount calculating step of the present invention. The CPU 1 functioning as the geometric feature quantity calculation unit 23 in S20, S32 and S35 in FIG. 7 corresponds to the feature quantity calculation means of the present invention. The process of S34 corresponds to the selection step of the present invention. The CPU 1 that executes the process of S34 functions as selection means of the present invention. The process of S37 corresponds to a reference point acquisition step of the present invention. CPU1 which performs the process of S37 functions as a reference point acquisition means of this invention. The process of S38 corresponds to the sample information acquisition step of the present invention. The CPU 1 that executes the process of S38 functions as sample information acquisition means of the present invention. The process of S40 corresponds to the frequency information acquisition step of the present invention. CPU1 which performs the process of S40 functions as a frequency information acquisition means of this invention. S42 and S4 correspond to the storage control step of the present invention. The CPU 1 that executes the processes of S42 and S4 functions as a storage control unit of the present invention. The process of S5 corresponds to the collation step of the present invention. The processing from S52 to S55 corresponds to the pair extraction step of the present invention. S58 to S62 and S68 correspond to the post-extraction collation step of the present invention.

  The information processing apparatus 10 that executes the information processing program of the present embodiment acquires sample information and frequency information based on the feature points extracted from the image (S38, S40). The sample information and the frequency information are information acquired for the second reference point determined based on the geometric feature amount represented by the three feature points extracted from the image. For example, based on the image 41 shown in FIG. 5, the sample information SDR1 to SDR3 in FIG. 11 is acquired for the three feature points P1, P4, and P8. Based on the image 51 shown in FIG. 14, the sample information SDT1 to SDT3 in FIG. 15 is acquired for the three feature points PT1, PT2, and PT3. The image 51 is an image taken with the biological information moved and rotated with respect to the image 41. Sample images 76 to 78 generated from the sample information SDR1 to SDR3 and sample images 86 to 88 generated from the sample information SDT1 to SDT3 represent the same striped pattern. That is, in the collation information of this embodiment, the influence of the biological information rotating or moving with respect to the reference is reduced. In this way, if the collation information is generated based on the information processing program, the biological information represented by the image is rotated with respect to the reference as compared to the case where the information is acquired about the point where the geometric feature amount is not considered. Thus, it is possible to generate collation information with reduced influence of movement.

  The sample information and frequency information acquired based on the three feature points are information obtained by efficiently and effectively acquiring the change pattern of the color information of the pixels around the two points of interest according to the conditions defined by the information processing program. is there. For example, if the second reference point is set on a square side that is arranged with the first reference point as a reference, the ridge and the side are arranged substantially in parallel depending on the direction in which the square side is arranged. May be. In such a case, even if the graphic in which the second reference point is arranged is changed from a circle to a square and the same processing as that of the present embodiment is performed, the CPU 1 does not display color information indicating characteristic biological information changes. Can not get. On the other hand, the CPU 1 sets the second reference point on the circumference of the circle having the center at the first reference point and the radius R, thereby changing the characteristic information of the color information of the pixels around the first reference point. Easy to get. For this reason, the information processing apparatus 10 can generate collation information that can be collated with high accuracy even when the number of feature points extracted from the image is relatively small.

  The sample image is an image in which color information included in the image is rearranged according to the conditions specified by the program, and can be said to represent real space information. On the other hand, the frequency image is obtained by converting real space information into frequency space information. For this reason, sample information and frequency information are considered to have different meanings. The information processing apparatus 10 according to the present embodiment generates collation information including both sample information and frequency information. Therefore, the amount of information can be increased compared to the case where either sample information or frequency information is included. In other words, by performing collation using the collation information of the present embodiment, it is appropriate whether or not it is similar from more viewpoints than when using information including only sample information or frequency information. Can be determined. In the present embodiment, after the sample information and the frequency information are standardized, the similarity D is calculated by multiplying the coefficient (contribution rate). For this reason, the similarity D can be obtained by relatively simple calculation using the sample information and the frequency information having different meanings.

  CPU1 of this embodiment acquires sample information and frequency information about three sets of two attention points corresponding to each of all combinations when two feature points are selected from three feature points. Compared to the case where a set of two attention points is selected, the number of sample information and frequency information to be acquired is increased, so that even when the number of feature points extracted from an image is small, it is possible to accurately collate Information can be generated.

  Since the information processing apparatus 10 collates biometric information using the collation information, collation can be performed with high accuracy even when the number of feature points extracted from the image is small. The sample information and the frequency information are information acquired for points determined based on the geometric feature amount. Therefore, the information processing apparatus 10 can reduce the influence that the biological information is rotated or moved with respect to the reference, compared with the case where information is acquired about the point where the geometric feature amount is not considered. Good matching results can be obtained.

  When the positions of the three feature points acquired in the feature point acquisition step are close to each other and the area of the triangle represented by the three feature points is less than a predetermined value, the area of the triangle is greater than or equal to the predetermined value Compared to the above, there is little change in the color information indicated by the sample information. For this reason, when the area of the triangle represented by the three feature points is less than the predetermined value, the information processing apparatus 10 does not acquire collation information based on the three feature points (S24). Therefore, the information processing apparatus 10 can increase the accuracy of the collation using the collation information as compared with the case where the collation information is generated based on the three feature points whose triangular area is less than the predetermined value.

  The CPU 1 can simplify the process of extracting a reference triangle similar to the test triangle by setting the coordinates of the local coordinate system. As a result, the CPU 1 can shorten the processing time required for collation.

  The CPU 1 of the present embodiment selects collation information used for collation, particularly in the case where the area is large. That is, the CPU 1 can further improve collation accuracy while shortening the processing time required for collation by preferentially using collation information with a large amount of information from collation information acquired based on images.

  The information processing program and information processing apparatus of the present invention are not limited to the above-described embodiments, and various changes may be made without departing from the scope of the present invention. For example, the following modifications (A) to (C) may be added as appropriate.

  (A) The configuration of the information processing apparatus 10 may be changed as appropriate. For example, the information processing apparatus 10 is not limited to a smartphone, and may be a mobile device such as a notebook PC, a tablet PC, and a mobile phone, or an automatic teller machine (ATM) and an input / output device. It may be a device such as an exit management device. The biological information acquisition device 8 may be provided separately from the information processing device 10. In that case, the biometric information acquisition apparatus 8 and the information processing apparatus 10 may be connected by a connection cable, or may be connected wirelessly such as Bluetooth (registered trademark) and NFC (Near Field Communication). Good. The detection method of the biological information acquisition device 8 is not limited to the capacitance method, and may be another method (for example, an electric field method, a pressure method, or an optical method). The biological information acquisition device 8 is not limited to a plane type, and may be a linear type. The size, color information, and resolution of the image generated by the biological information acquisition apparatus 8 may be changed as appropriate. Therefore, for example, the color information may be information corresponding to a color image in addition to information corresponding to a black and white image.

  (B) The information processing program may be stored in the storage device of the information processing apparatus 10 before the information processing apparatus 10 executes the program. Therefore, each of the information processing program acquisition method, the acquisition route, and the device storing the information processing program may be changed as appropriate. The information processing program executed by the processor of the information processing device 10 may be received from another device via a cable or wireless communication and stored in a storage device such as a flash memory. Other devices include, for example, a PC (personal computer) and a server connected via a network.

  (C) Each step of the collation information processing in FIG. 3 is not limited to the example executed by the CPU 1, and a part or all of the steps may be executed by another electronic device (for example, ASIC). Each step of the above process may be distributedly processed by a plurality of electronic devices (for example, a plurality of CPUs). Each step of the collation information processing of the above embodiment can be changed in order, omitted or added as necessary. Based on a command from the CPU 1 of the information processing apparatus 10, an operating system (OS) or the like operating on the information processing apparatus 10 performs part or all of the actual processing, and the functions of the above embodiments are realized by the processing. Such cases are also included in the scope of the present disclosure. For example, the following changes (C-1) to (C-6) may be appropriately added to the verification information processing.

  (C-1) Preprocessing may be appropriately performed on the image acquired in S11. For example, a filtering process for removing high frequency components of an image as noise may be executed. By executing the filtering process, the change in shading of the edge portion of the image is moderated. As a filter used for the filtering process, any of a known low-pass filter, Gaussian filter, moving average filter, median filter, and averaging filter may be used. In another example, a filtering process for extracting only a specific frequency band component may be performed on the image acquired in S11. As a specific frequency band, a band including the period of the concave and convex portions of the fingerprint may be selected. In this case, a known band-pass filter may be used as a filter used for the filtering process.

  (C-2) The conditions for extracting the three feature points that are the targets for acquiring the collation information in S21 of FIG. 4 may be changed as appropriate. For example, the CPU 1 may determine whether to acquire collation information based on whether the value of the maximum inner angle is within a predetermined range in addition to the area of the triangle formed by the three points. CPU1 may abbreviate | omit the process of S21 and may acquire collation information about each of the three feature points corresponding to all the combinations calculated by S18. CPU1 may acquire collation information about a predetermined set of three feature points in descending order of area.

  (C-3) In S58, the frequency information similarity calculation method may be changed as appropriate. For example, when a one-dimensional group delay spectrum similar to that of the present embodiment is used as a frequency component, a noise component may appear strongly in a higher-order component. In consideration of such a case, the frequency information may be selected based on frequency information including a predetermined number of components selected with priority given to the low-order components. The predetermined number may be determined in advance in consideration of the number of samples and the authentication accuracy. For example, when the number N of samples acquired for one first reference point is 128, the predetermined number is 10 to 63. Set to either. Preferably, the predetermined number is set to any one of 12 to 20. In the case of the sample number N, the predetermined number is preferably set from (sample number N / 10) to (sample number N / 5).

  (C-4) The frequency component is not limited to the one-dimensional group delay spectrum. For example, other known frequency components such as an LPC spectrum, a group delay spectrum, an LPC cepstrum, a cepstrum, an autocorrelation function, and a cross correlation function may be used as the frequency component.

  (C-5) Collation may be executed in combination with known collation information. For example, the final determination may be performed by combining a matching result by a known minutiae method and a matching result using the matching information method of the present invention. In this way, it is expected that collation is performed from various viewpoints and collation accuracy is improved.

  (C-6) The setting conditions (for example, the number, interval, acquisition order, etc.) of the first reference point and the second reference point may be changed as appropriate. Sample information and frequency information may be acquired for one side or two sides of the three sides of the triangle. One direction vector of S35 may be calculated for one triangle. In this case, when sample information and frequency information are acquired for a plurality of sides among the three sides of the triangle, the start point SP may be determined based on a common direction vector. The three pieces of sample information calculated for each of the three sides of the triangle may be appropriately combined and handled as one piece of sample information. Similarly, the three frequency information calculated for each of the three sides of the triangle may be appropriately combined and handled as one frequency information. In this case, it is possible to collate using all of the three sample information and the three frequency information calculated for each of the three sides. The process of generating the sample image and the process of generating the frequency image may be omitted as appropriate. The geometric feature amount, the first information, and the second information may be changed as appropriate.

1 CPU
2 ROM
3 RAM
4 Flash memory 10 Information processing device 21 Image acquisition unit 22 Feature point extraction unit 23 Geometric feature quantity calculation unit 24 Sample information acquisition unit 25 Frequency information acquisition unit 26 Registration unit 27 Collation unit 28 DB

Claims (5)

In a computer having a processor and storage means,
An image acquisition step of acquiring an image;
Extracting the feature point by analyzing the image acquired in the image acquisition step;
A feature point acquisition step of acquiring three of the feature points extracted in the extraction step;
A feature amount calculating step of calculating a geometric feature amount of a triangle represented by the three feature points acquired in the feature point acquiring step;
A selection step of selecting two feature points as attention points from among the three feature points acquired in the feature point acquisition step;
First information corresponding to a position of a first reference point, which is a point on a line segment connecting the two points of interest selected in the selection step and determined based on the geometric feature amount. A reference point obtaining step for obtaining each of the plurality of first reference points having different positions from each other;
For each of the plurality of first information acquired in the reference point acquisition step,
A point on a circumference of a circle having a predetermined radius centered on the first reference point corresponding to the first information, and corresponding to a second reference point determined based on the geometric feature amount Samples that are information in which color information and second information corresponding to the position of the second reference point are associated with each other are acquired for each of the plurality of second reference points having different positions,
A sample information acquisition step for performing processing in which sample information is obtained by associating a plurality of acquired samples with the first information;
Based on the sample information acquired in the sample information acquisition step, for each of the plurality of first reference points,
Calculating a frequency component of the color information change with respect to the second information;
A frequency information acquisition step of acquiring frequency information, which is information in which the frequency component calculated for each of the plurality of first reference points is associated with the first information;
An information processing program for executing a storage control step of storing, in the storage unit, information in which the sample information and the frequency information are associated with the geometric feature amount as verification information used for verification of biological information.   2. The information processing program according to claim 1, wherein the selection step selects each of three sets of the two attention points corresponding to a combination of all two of the three feature points.   The information processing program according to claim 1, further comprising a collation step of collating the collation information for testing to be collated with the collation information for reference stored in the storage unit. . The feature point acquisition step includes a case where the three feature points are selected from a plurality of feature points within a predetermined range determined based on the predetermined radius among the plurality of feature points extracted in the extraction step. , Obtain a plurality of sets of the three feature points corresponding to all combinations,
In the selection step, an area of the triangle is a predetermined value based on the geometric feature amount calculated in the feature amount calculation step among the plurality of sets of three feature points acquired in the feature point acquisition step. For the above three feature points, a process of selecting the two attention points is executed,
The matching step includes
The geometric feature amount of each of the test triangle represented by the test collation information and the reference triangle represented by the reference collation information is large in the area of the triangle. A pair extraction step of extracting a pair of the test triangle and the reference triangle, the similarity of which is compared in order, and the similarity is equal to or less than a threshold;
A post-extraction collation step of collating the sample information and the frequency information corresponding to the test triangle extracted in the pair extraction step with the sample information and the frequency information corresponding to the reference triangle. The information processing program according to claim 3, further comprising: Storage means for storing information;
Image acquisition means for acquiring images;
Analyzing the image acquired by the image acquisition means, extracting the feature points;
Feature point acquisition means for acquiring three feature points extracted by the extraction means;
Feature amount calculating means for calculating a geometric feature amount of a triangle represented by the three feature points acquired by the feature point acquiring means;
Selection means for selecting two feature points as attention points from among the three feature points acquired by the feature point acquisition means;
First information corresponding to a position of a first reference point, which is a point on a line segment connecting the two points of interest selected by the selection means and determined based on the geometric feature amount, Reference point acquisition means for acquiring each of the plurality of first reference points whose positions are different from each other;
For each of the plurality of first information acquired by the reference point acquisition means,
A point on a circumference of a circle having a predetermined radius centered on the first reference point corresponding to the first information, and corresponding to a second reference point determined based on the geometric feature amount Samples that are information in which color information and second information corresponding to the position of the second reference point are associated with each other are acquired for each of the plurality of second reference points having different positions,
Sample information acquisition means for performing a process using the information obtained by associating the plurality of acquired samples with the first information as sample information;
Based on the sample information acquired by the sample information acquisition means, for each of the plurality of first reference points,
Calculating a frequency component of the color information change with respect to the second information;
Frequency information acquisition means for acquiring frequency information which is information in which the frequency component calculated for each of the plurality of first reference points is associated with the first information;
Storage control means for storing information in which the sample information and the frequency information are associated with the geometric feature amount in the storage means as collation information used for collation of biometric information. apparatus.