JP4561169B2 - Image collation device, image collation processing program, and image collation method - Google Patents

Description

  The present invention relates to an image collation apparatus, an image collation processing program, and an image collation method for collating fingerprint image data, for example.

  In recent years, for example, in an entrance / exit management system for an important facility or an access permission system for various information devices, a system for performing personal authentication based on a biometric image such as a fingerprint image, an iris image, or a face image has been widely introduced. It has become.

  In such a personal authentication system, for example, in the case of a fingerprint image collation device, first, a fingerprint image of a specific finger (for example, an index finger) of a regular person read by an image sensor or the like is stored in advance as a registered fingerprint image. . Then, by using the fingerprint image of the specific finger of the target person read in the same manner for the personal authentication as a collation fingerprint image, the collation similarity with the registered fingerprint image stored in advance is calculated, so that the registered fingerprint image and the collation fingerprint The identity with the image is determined, and whether or not the person is a registered person is authenticated.

  On the other hand, for fingerprints, the fingerprint pattern itself is basically the same pattern even with age, but due to changes in body shape such as aging and deterioration of the skin, fatness and thinning, etc. The fingerprint image read by the image sensor undergoes changes such as the crest of the fingerprint pattern being crushed with time. In particular, when the finger is a dominant finger such as an index finger, the fingerprint surface is rubbed and the change may be remarkable.

  For this reason, if the registered fingerprint image is an initially registered fingerprint image, even if it is a collation fingerprint image obtained by reading the fingerprint of the registered person after the change over time, it is not determined to be the same fingerprint and is difficult to collate Become.

Therefore, when the collation fingerprint image matches the registered fingerprint image and the authentication of the registered person is obtained, the collation fingerprint image at that time is stored and updated as a new registered fingerprint image each time the collation coincides. Thus, a fingerprint image collation apparatus that prevents an erroneous determination such as rejection of the person due to a change with time has been considered (see, for example, Patent Document 1).
Japanese Patent Laid-Open No. 11-25268

  However, each time a collation fingerprint image collates with the registered fingerprint image, the conventional fingerprint image collation apparatus in which the collation fingerprint image at that time is stored and updated as a new registered fingerprint image is registered relatively frequently. In a state where fingerprint verification is performed by the person himself, there is little amount of fingerprint image displacement due to changes in the verification fingerprint over time and finger placement, and smooth fingerprint verification processing and registration fingerprint image update processing are repeated, If the finger placement changes or you forget to place it rarely, and the registration fingerprint image matches and matches the registered fingerprint image while the amount of misalignment of the verification fingerprint image is large, the registration will be verified in subsequent verification processing. It takes time-consuming processing to make a match with the registered fingerprint image while shifting the fingerprint image little by little. There is that problem.

The present invention has been made in view of the above problems. For example, even when the collation fingerprint image and the registered fingerprint image match, the registration fingerprint image is updated when the amount of deviation due to the placement of the collation fingerprint image is large. An image collation apparatus and an image collation process capable of preventing bad influence on the subsequent collation processing without performing it and enabling good collation even when the finger placement is changed habitually An object is to provide a program and an image collation method.

An image collating apparatus according to the present invention collates a registered image and a collated image and determines whether or not they match, and the image collating unit determines that the registered image and the collated image match. A displacement amount calculating means for calculating a displacement amount of the collation image with respect to the registered image, and a position displacement threshold value in which the displacement amount of the collation image with respect to the registered image calculated by the displacement amount calculation means is set in advance. The registration update unit that updates the collation image as a new registration image that replaces the previous registration image, and the registration collation image and the collation image each time the registration image is determined to match. A deviation frequency distribution storage means for storing a positional deviation amount of the collation image with respect to the registered image calculated by an amount calculation means as a positional deviation frequency distribution; and a deviation amount calculation means. Even when the calculated positional deviation amount of the collation image with respect to the registered image is equal to or larger than a predetermined positional deviation threshold, the positional deviation frequency stored by the deviation frequency distribution storage means corresponding to the positional deviation amount. It is characterized by comprising high-registration frequency registration update means for updating the collation image as a new registered image that replaces the previous registered image when the frequency exceeds a preset frequency threshold.

According to the present invention, even when the registered image matches the collated image and is confirmed as a normally registered collated image, if the amount of positional deviation of the collated image is large, the registered image is not updated and the subsequent images are updated. It is possible to prevent the collation process from being adversely affected, and it is possible to perform a good collation because the registered image is updated accordingly when the manner of placing the finger is changed habitually .

  Embodiments of the present invention will be described below with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram showing a configuration of an electronic circuit of a fingerprint image matching apparatus according to an embodiment of an image matching apparatus of the present invention.

  This fingerprint image collation device is composed of a control unit (CPU) 11, a storage device 12, a RAM 13, an image reading device 14, a display unit 15, an input unit 16, a storage medium reading unit 18, and a transmission control unit 19. Are connected to each other.

  In response to an input signal from the input unit 12, the control unit (CPU) 11 is a control program stored in advance in the storage device 12, or a storage medium such as a floppy disk drive FDD from an external storage medium 18a such as a floppy disk FD. The fingerprint collation processing program read into the storage device 12 via the reading unit 18 or the storage device 22 of the external computer terminal (program server 21) from the storage device 12 via the communication network 20 to the storage device 12 via the transmission control unit 19. The read fingerprint collation processing program is activated, and the operation of each part of the circuit is controlled using the RAM 13 as a work memory.

  The storage device 12 is constituted by a ROM of a semiconductor memory, a hard disk device HDD, or the like. In addition to storing the above-described control program read by the control unit (CPU) 11 immediately after turning on the power of the verification device, the storage device 12 reads an image. The registered fingerprint image Go read by the device 14 is stored.

  The RAM 13 is a work memory used by the control unit (CPU) 11 when executing the aforementioned control program.

  The image reading device 14 is a device for acquiring a registered fingerprint image Go and a collation fingerprint image Gs, and is an image sensor such as an image scanner device or a CCD (charge coupled device).

  The display unit 15 is a display device such as a CRT or a liquid crystal display that displays a result of collation between the registered fingerprint image Go and the collation fingerprint image Gs.

  The input unit 16 is an input device such as a keyboard device for the user of the verification device to instruct the control unit (CPU) 11 to issue a fingerprint image acquisition instruction or start a fingerprint verification process described later.

  FIG. 2 is a diagram showing a data memory secured in the RAM 13 of the fingerprint image collating apparatus.

  The RAM 13 includes a read image memory 13a, a registered fingerprint image memory 13b, a verification fingerprint image memory 13c, a reference mask size / position information memory 13d, a maximum correlation position information memory 13e, a verification determination result memory 13f, a horizontal shift amount memory 13g, A vertical deviation amount memory 13h, a horizontal deviation frequency distribution memory 13i, a vertical deviation frequency distribution memory 13j, a rotation direction deviation amount memory 13k, a rotation deviation frequency distribution memory 13l, and the like are secured.

  In the read image memory 13a, fingerprint image data (see FIG. 6) read by the image reading device 14 is temporarily stored as a multi-tone image.

  The registered fingerprint image memory 13b stores and holds the fingerprint image data Go (see FIG. 6A) of the registered person read by the image reading device 14 to be stored and registered in the storage device 12.

  In the collation fingerprint image memory 13c, fingerprint image data Gs (see FIG. 6B) to be collated read by the image reading device 14 is stored as a multi-tone image.

  In the reference mask size / position information memory 13d, reference mask regions (templates) t1, t2,..., Tn arranged on one fingerprint image data (Go or Gs) to be verified in the fingerprint verification apparatus (this embodiment) In this case, the mask size and arrangement position information of t1 and t2 (see FIG. 6A) are set and stored in advance by default.

  The maximum correlation position information memory 13e stores the other fingerprint image data (Gs or Go) for each image of the reference mask areas t1 and t2 arranged on the one fingerprint image data (Go or Gs) to be verified. The position information of each maximum correlation (position) region k1, k2 (see FIG. 6B) where the maximum correlation coefficient is calculated is searched and stored.

  The collation determination result memory 13f stores the mutual positional relationship XYt (see FIG. 6A) of the reference mask areas t1 and t2 with respect to the one fingerprint image data (Go or Gs) and the other fingerprint image data. The registered fingerprint image Go and the verification fingerprint image Gs that are evaluated according to the difference from the mutual positional relationship XYk (see FIG. 6B) of the maximum correlation (position) regions k1 and k2 on (Gs or Go) Match / mismatch determination results are stored.

  In the horizontal direction displacement amount memory 13g, the finger on the image reading device 14 of the verification fingerprint image data Gs stored in the verification fingerprint image memory 13c with respect to the registered fingerprint image data Go stored in the registration fingerprint image memory 13b is stored. The amount of horizontal translational deviation (ΔH) resulting from the placement of the image is, for example, the amount of horizontal deviation of the center of gravity between the fingerprint images Go and Gs, or the reference mask areas t1 and t2 arranged on the registered fingerprint image data Go. Corresponding to this, it is calculated and stored as a horizontal direction shift amount with respect to the maximum correlation (position) regions k1 and k2 searched on the collation fingerprint image data Gs.

  In the vertical direction displacement amount memory 13h, the finger on the image reading device 14 of the verification fingerprint image data Gs stored in the verification fingerprint image memory 13c with respect to the registered fingerprint image data Go stored in the registration fingerprint image memory 13b is stored. The amount of vertical translational deviation (ΔV) resulting from the placement of the image is, for example, the amount of vertical deviation of the center of gravity between the fingerprint images Go and Gs, or the reference mask areas t1 and t2 arranged on the registered fingerprint image data Go. Corresponding to this, it is calculated and stored as a vertical direction deviation amount with respect to the maximum correlation (position) regions k1 and k2 searched on the collation fingerprint image data Gs.

  The horizontal shift frequency distribution memory 13i stores the horizontal translation shift amount (ΔH) calculated every time the fingerprint registration processing is repeatedly performed on the same registered fingerprint image data Go with the verification fingerprint image data Gs. The frequency distribution (Histogram) [ΔH] is calculated and accumulated and stored.

  FIG. 3A is a diagram showing frequency distribution (Histogram) data of the horizontal translational deviation (ΔH) stored in the horizontal deviation frequency distribution memory 13i of the fingerprint image matching device.

  According to the frequency distribution data of the amount of horizontal translational deviation, how much horizontal translational deviation is caused by the placement of the finger of the person to be collated at the time of fingerprint collation performed repeatedly on the registered fingerprint image data Go. (Hn) It is possible to determine how often it occurs, and this horizontal shift frequency distribution data includes a frequency threshold Th− for determining that horizontal shift by the person to be collated is habitual. In addition to F being set, a horizontal shift allowable threshold Th-H is set for determining the upper limit of the horizontal shift Hn that can be allowed in order not to cause erroneous determination such as rejection of the person or delay of the collation processing time.

  In the vertical deviation frequency distribution memory 13j, the amount of vertical translation deviation (ΔV) calculated every time the fingerprint collation processing with the repeated fingerprint image data Gs is repeatedly performed on the same registered fingerprint image data Go. The frequency distribution (Histogram) [ΔH] is calculated and accumulated and stored.

  FIG. 3B is a diagram showing frequency distribution (Histogram) data of the vertical translational deviation amount (ΔV) stored in the vertical deviation frequency distribution memory 13j of the fingerprint image matching device.

  According to the frequency distribution data of the amount of vertical translational deviation, how much vertical translational deviation is caused by how to place the finger of the person to be collated at the time of fingerprint collation performed repeatedly on the registered fingerprint image data Go. (Vn) It is possible to determine how often it occurs, and in this vertical deviation frequency distribution data, a frequency threshold Th− for determining that horizontal deviation by the person to be collated is customary. In addition to F being set, a vertical deviation allowable threshold Th-V is set for determining an upper limit of the vertical deviation Vn that can be tolerated in order not to cause erroneous determination such as refusal of the person or a delay in collation processing time.

  In the rotational direction deviation amount memory 13k, a finger on the image reading device 14 of the verification fingerprint image data Gs stored in the verification fingerprint image memory 13c with respect to the registered fingerprint image data Go stored in the registration fingerprint image memory 13b is stored. The amount of deviation (ΔR) in the rotation direction due to the placement of the image is, for example, the amount of vertical deviation of the center of gravity position between the fingerprint images Go and Gs, or the reference mask regions t1 and t2 arranged on the registered fingerprint image data Go. Corresponding to this, it is calculated and stored as a rotational direction shift amount with respect to the maximum correlation (position) regions k1 and k2 searched on the collation fingerprint image data Gs.

  The rotation deviation frequency distribution memory 13l stores the rotation direction deviation amount (ΔR) calculated each time the fingerprint registration process is repeatedly performed on the same registered fingerprint image data Go with the fingerprint image data Gs. Calculated and stored as a frequency distribution (Histogram) [ΔR].

  According to the frequency distribution data of the rotational direction deviation amount, the degree of rotational direction deviation (Rn) is caused by how to put the finger of the person to be collated at the time of fingerprint collation performed repeatedly on the registered fingerprint image data Go. ) It is possible to determine the frequency of occurrence, and for this rotational deviation frequency distribution data, as in the case of the horizontal deviation (Hn) and vertical deviation (Vn) frequency distribution data, A frequency threshold Th-F for determining that the rotation deviation by the user is habitual is set, and the allowable rotation deviation Rn is avoided in order not to cause misjudgment such as refusal of the person or delay of the verification processing time. A rotation deviation allowable threshold value Th-R for determining the upper limit of is set.

  In the fingerprint image collation apparatus according to the first embodiment, when the collation match determination between the registered fingerprint image data Go and the collation fingerprint image data Gs is obtained, the horizontal direction corresponding to the finger placement of the collation target person at that time is obtained. It is confirmed that both the direction translational deviation amount ΔH and the vertical direction translational deviation amount ΔV are less than the permissible threshold values Th-H and TH-V, and the matching fingerprint image data Gs determined to match this time is registered fingerprint image data Go. The registered fingerprint image data Go is updated at any time according to the time-dependent change of the registered person's fingerprint, and the collated fingerprint image data Gs having a large deviation in the fingerprint collection position is transferred to the registered fingerprint image data Go. Is not updated, and it is possible to prevent the possibility of erroneous determination such as refusal of identity and delay of verification processing time in subsequent fingerprint verification.

  Next, the fingerprint collation / update function of the first embodiment in the fingerprint image collation apparatus having the above configuration will be described.

  FIG. 4 is a flowchart showing fingerprint collation / update processing of the first embodiment in the fingerprint image collation apparatus.

  FIG. 5 is a flowchart separately showing a fingerprint collation process accompanying the fingerprint collation / update process of the fingerprint image collation apparatus.

  FIG. 6 shows the arrangement state of the reference mask regions (templates) t1 and t2 on the registered fingerprint image Go and the maximum correlation position (region) k1 on the verification fingerprint image Gs in accordance with the fingerprint verification processing of the fingerprint image verification device. It is a figure which shows k2.

  In the fingerprint collation / update process of the first embodiment shown in FIG. 4, first, the fingerprint collation process shown in detail in FIG. 5 is started (step AS).

  In this fingerprint collation processing, when the fingerprint image of the person to be collated is read by the image reading device 14 and stored in the read image memory 13a in the RAM 13, the read image is compared with the collation fingerprint image data Gs to the collation fingerprint image memory 13c. (Step S1).

  Then, as shown in FIG. 6A, for the registered fingerprint image data Go stored in the registered fingerprint image memory 13b, the reference mask information stored and defined in the reference mask size / position information memory 13d is added. Based on this, two reference mask areas t1 and t2 based on the position of the center of gravity of the registered fingerprint image Go are arranged (step S2).

  Then, each maximum correlation (the maximum correlation coefficient is calculated on each of the two reference mask areas t1 and t2 on the registered fingerprint image Go and the collation fingerprint image Gs stored in the collation fingerprint image memory 13c). Position) regions k1 and k2 are searched for each reference mask as shown in FIG. 6B (steps S3 and S4).

  Then, the positional relationship XYt between the reference masks (regions) t1 and t2 arranged in the registered fingerprint image Go and the positional relationship XYk between the maximum correlation (position) regions k1 and k2 detected on the collation fingerprint image Gs are obtained. The degree of similarity of each positional relationship XYt: XYk is determined (steps S5 and S6).

  The degree of similarity (positional deviation) according to an error (positional deviation) between the positional relationship XYt between the reference masks (areas) t1 and t2 and the positional relation XYk between the maximum correlation (position) areas k1 and k2 It is determined whether or not “Small” → Similarity “Large”) is greater than or equal to a preset threshold value (step S 7). This is done (step S9).

  Then, the matching result of the matching fingerprint image Gs with respect to the registered fingerprint image Go is stored in the matching determination result memory 13f. For example, when the fingerprint image matching device is used in an entrance / exit management system, Entrance / exit is permitted / rejected according to the determination result of the match / mismatch.

  When the fingerprint collation processing of the collation fingerprint image data Gs with respect to the registered fingerprint image data Go is performed (step AS), if it is determined that the person matches (step A1), the collation fingerprint for the registration fingerprint image data Go is determined. The horizontal and vertical translational deviation amounts (ΔH, ΔV) caused by the finger placement of the image Gs are, for example, the horizontal and vertical deviations of the center of gravity position between the fingerprint images Go and Gs, or the registered fingerprint image Go. Is calculated as a horizontal and vertical shift amount between the reference mask areas t1 and t2 arranged in the above and the maximum correlation (position) areas k1 and k2 searched on the matching fingerprint image Gs corresponding to the reference mask areas t1 and t2. 13g and the vertical deviation memory 13h are stored (step A2).

  Whether the horizontal translation deviation amount ΔH and the vertical translation deviation amount ΔV stored in the horizontal deviation memory 13g and the vertical deviation memory 13h are less than the allowable threshold values Th-H and TH-V. It is determined whether or not (step A3).

  Here, it is assumed that the horizontal translational deviation amount ΔH and the vertical translational deviation amount ΔV of the collation fingerprint image Gs that has been collated and matched this time with the registered fingerprint image Go are both less than the allowable threshold values Th-H and TH-V. If it is determined, the verification fingerprint image data Gs is overwritten and stored in the registered fingerprint image memory 13b as registered fingerprint image data Go (step A3 → A4).

  On the other hand, if it is determined that any one of the horizontal translational deviation amount ΔH and the vertical translational deviation amount ΔV is not less than the allowable threshold values Th-H and TH-V, Even if the collation fingerprint image data Gs is confirmed, the series of fingerprint collation / update processing is terminated without performing the overwrite update processing on the registered fingerprint image data Go (step A3 → end).

  Therefore, according to the fingerprint collation / update function of the first embodiment in the fingerprint image collation apparatus having the above-described configuration, when a collation coincidence determination between the registered fingerprint image Go and the collation fingerprint image Gs is obtained, When the horizontal translational deviation amount ΔH and the vertical translational deviation amount ΔV corresponding to the placement are both less than the allowable threshold values Th-H and TH-V, the matching fingerprint image Gs determined to match this time is registered fingerprint. If the image is updated as the image Go and does not fall below the allowable threshold values Th-H and TH-V, the registered fingerprint image Go is not updated. While the registered fingerprint image Go is updated at any time, the verification fingerprint image Gs in which the fingerprint collection position has greatly shifted is not updated to the registered fingerprint image Go. Judgment Kill is possible to prevent the possibility of causing a delay matching the processing time in advance.

  In the fingerprint collation / update function of the first embodiment, the threshold value Th is acceptable for either the horizontal translational deviation amount ΔH or the vertical translational deviation amount ΔV of the collated fingerprint image Gs that has been collated with the registered fingerprint image Go. When it is determined that it does not fall below -H, TH-V, the registered fingerprint image Go is not updated unconditionally. However, in the fingerprint collation / update function of the second embodiment, As will be described, even when it is determined that the horizontal and vertical translational deviation amounts (ΔH, ΔV) of the verification fingerprint image Gs are not less than the allowable threshold values Th-H, TH-V, the large deviation amount ( If the frequency distribution [ΔH] [ΔV] (see FIG. 3) corresponding to (ΔH) (ΔV) is equal to or higher than the frequency threshold Th-F indicating that it is habitual, future verification processing But the same horizontal shift ΔH and vertical Assuming that a collation fingerprint image Gs with ΔV is obtained, the collation fingerprint image Gs having a large shift amount (ΔH) (ΔV) that is greater than or equal to the frequency threshold Th-F is updated as the registered fingerprint image Go. It is good also as a structure.

  As a result, not only the registered fingerprint image Go is updated at any time according to the time-dependent change of the registered person's fingerprint, but also updated when the finger placement associated with the person's fingerprint collation has customarily changed. .

(Second Embodiment)
FIG. 7 is a flowchart showing fingerprint collation / update processing of the second embodiment in the fingerprint image collation apparatus.

  In the fingerprint collation / update process of the second embodiment, the same processing steps as those of the fingerprint collation / update process of the first embodiment in FIG. 4 are denoted by the same step symbols and description thereof is omitted.

  That is, in the fingerprint collation / update process according to the second embodiment, when it is determined that the person matches according to the fingerprint collation process (step AS) (step A1), the horizontal and vertical directions of the collation fingerprint image Gs with respect to the registered fingerprint image Go. Are calculated and stored in the horizontal direction displacement amount memory 13g and the vertical direction displacement amount memory 13h (step A2), and the horizontal and vertical direction translational displacement amounts (ΔH, ΔV) are also calculated. ) Each frequency distribution [ΔH, ΔV] (see FIG. 3) is calculated and stored in the horizontal shift frequency distribution memory 13i and the vertical shift frequency distribution memory 13j (step A2a).

  Then, it is determined that both the horizontal translational deviation amount ΔH and the vertical translational deviation amount ΔV of the collated fingerprint image Gs that has been collated and matched this time with respect to the registered fingerprint image Go are less than the allowable threshold values Th−H and TH−V. In the case where it has been done, as in the first embodiment, the collated fingerprint image data Gs is overwritten and stored in the registered fingerprint image memory 13b as registered fingerprint image data Go (step A3 → A4).

  On the other hand, if it is determined that any one of the horizontal translational deviation amount ΔH and the vertical translational deviation amount ΔV is not less than the allowable threshold values Th−H and TH−V, the allowable threshold value Th−. A frequency threshold Th indicating that the frequency distribution [ΔH] or [ΔV] (see FIG. 3) corresponding to a large shift amount (ΔH or ΔV) that is not less than H or TH−V is customary. It is determined whether or not −F or more (step A3 → A5).

  Even when it is determined that any one of the horizontal and vertical translational deviation amounts (ΔH, ΔV) of the verification fingerprint image Gs does not fall below the allowable threshold values Th-H, TH-V, If it is determined that the frequency distribution [ΔH] or [ΔV] corresponding to the shift amount (ΔH or ΔV) is equal to or higher than the frequency threshold Th-F indicating that it is habitual, future verification It is predicted that the collation fingerprint image Gs having the same horizontal deviation ΔH and vertical deviation ΔV can be obtained even in the processing, and the collation fingerprint having a large deviation amount (ΔH) (ΔV) that is equal to or greater than the frequency threshold Th-F. The image Gs is overwritten and stored in the registered fingerprint image memory 13b as a registered fingerprint image Go (step A5 → A6).

  Then, the horizontal deviation frequency distribution data and the vertical deviation frequency distribution data so far stored in the horizontal deviation frequency distribution memory 13i and the vertical deviation frequency distribution memory 13j are cleared and initialized (step A7).

  On the other hand, if it is determined that any one of the horizontal and vertical translational deviation amounts (ΔH, ΔV) of the verification fingerprint image Gs does not fall below the allowable thresholds Th-H, TH-V, When it is determined that the frequency distribution [ΔH] or [ΔV] corresponding to the deviation amount (ΔH or ΔV) is not equal to or higher than the frequency threshold Th-F indicating that it is habitual, Even if the collation fingerprint image Gs is confirmed to be the identity of the registered person, the above-described series of fingerprint collation / update processing is terminated without performing overwriting update processing on the registered fingerprint image Go (step A5 → end).

  Therefore, according to the fingerprint collation / update function of the second embodiment of the fingerprint image collation apparatus having the above-described configuration, when a collation match determination between the registered fingerprint image Go and the collation fingerprint image Gs is obtained, When the horizontal translational deviation amount ΔH and the vertical translational deviation amount ΔV corresponding to the placement are both less than the allowable threshold values Th-H and TH-V, the matching fingerprint image Gs determined to match this time is registered fingerprint. Even when it is updated as the image Go and does not fall below the allowable threshold values Th-H and TH-V, the frequency distribution [ΔH] or [ΔV] corresponding to the large deviation amount ΔH or ΔV becomes customary. If the frequency threshold Th-F is greater than or equal to the threshold value Th-F, it is determined that a collation fingerprint image Gs having the same horizontal deviation ΔH and vertical deviation ΔV can be obtained in the future collation processing. Less than frequency threshold Th-F Since the verification fingerprint image Gs having the large deviation amount ΔH or ΔV is updated as the registered fingerprint image Go, it is only necessary to update the registered fingerprint image Go according to the temporal change of the registered person's fingerprint. In the case of a collation fingerprint image Gs that has rarely shifted significantly, it is not updated to the registered fingerprint image Go, and the registration is updated when the finger placement associated with the fingerprint collation of the registered person has customarily changed. In the subsequent fingerprint collation processing, it is possible to prevent the possibility of erroneous determination such as refusal of the identity and the delay of the collation processing time.

  In the fingerprint collation / update function of the first embodiment and the second embodiment, the translational displacement amount in the horizontal direction between the registered fingerprint image Go and the collation fingerprint image Gs determined to collate in any of the embodiments. When ΔH and the vertical translational deviation amount ΔV are less than the allowable threshold values Th-H and TH-V, the registered fingerprint image Go is unconditionally updated. However, the fingerprint of the following third embodiment is used. As described in the verification / update function, the frequency distribution [ΔH] [ΔV] (see FIG. 3) corresponding to the horizontal and vertical translational shift amounts (ΔH, ΔV) of the verification fingerprint image Gs at that time is customary. If the frequency threshold Th-F is higher than the target threshold value Th-F, the collation fingerprint image Gs having the same horizontal deviation ΔH and vertical deviation ΔV can be obtained in the future collation processing. More than this frequency threshold Th-F Along with the update in the case where the collation fingerprint image Gs having a large deviation amount (ΔH) (ΔV) is updated as the registered fingerprint image Go, and the manner of placing the finger accompanying the fingerprint collation of the registered person himself / herself has customarily changed, A configuration may also be adopted in which the update is performed in accordance with the temporal change of the personal fingerprint.

(Third embodiment)
FIG. 8 is a flowchart showing fingerprint collation / update processing of the third embodiment in the fingerprint image collation apparatus.

  In the fingerprint collation / update process of the third embodiment, the same processing steps as those of the fingerprint collation / update process of the second embodiment in FIG. 2 are denoted by the same step symbols, and the description thereof is omitted.

  That is, in the fingerprint collation / update process according to the third embodiment, when it is determined that the person matches according to the fingerprint collation process (step AS) (step A1), the horizontal and vertical directions of the collation fingerprint image Gs with respect to the registered fingerprint image Go. Are calculated and stored in the horizontal direction displacement amount memory 13g and the vertical direction displacement amount memory 13h (step A2), and the horizontal and vertical direction translational displacement amounts (ΔH, ΔV) are also calculated. ) Each frequency distribution [ΔH, ΔV] (see FIG. 3) is calculated and stored in the horizontal shift frequency distribution memory 13i and the vertical shift frequency distribution memory 13j (step A2a).

  The frequency distribution [ΔH] [ΔV] corresponding to the horizontal and vertical translational deviation amounts (ΔH, ΔV) of the verification fingerprint image Gs is not less than a frequency threshold Th-F indicating that it is customary. If it is determined that the collation fingerprint image Gs having the same level of horizontal deviation ΔH or vertical deviation ΔV is obtained in the future collation processing, the deviation that exceeds the frequency threshold Th−F is predicted. The verification fingerprint image Gs having the amount (ΔH) (ΔV) is overwritten and stored in the registered fingerprint image memory 13b as the registered fingerprint image Go (step A5a → A6).

  Then, the horizontal deviation frequency distribution data and the vertical deviation frequency distribution data so far stored in the horizontal deviation frequency distribution memory 13i and the vertical deviation frequency distribution memory 13j are cleared and initialized (step A7).

  On the other hand, the frequency distribution [ΔH] [ΔV] corresponding to the horizontal and vertical translational deviation amounts (ΔH, ΔV) of the verification fingerprint image Gs is higher than the frequency threshold Th-F indicating that it is customary. If it is determined that the registered fingerprint image Gs has been verified, the series of fingerprint verification / update processing is performed without performing the update update processing on the registered fingerprint image Go even if the verification fingerprint image Gs is confirmed to be the identity of the registered person. The process is terminated (Step A5a → End).

  Therefore, according to the fingerprint collation / update function of the third embodiment in the fingerprint image collation apparatus having the above-described configuration, when the collation coincidence determination between the registered fingerprint image Go and the collation fingerprint image Gs is obtained, The frequency distribution [ΔH] [ΔV] corresponding to the horizontal translational deviation amount ΔH and the vertical translational deviation amount ΔV according to the placement is equal to or higher than the frequency threshold Th-F indicating that it is customary. In this case, it is predicted that a collation fingerprint image Gs having the same level of horizontal deviation ΔH and vertical deviation ΔV will be obtained in future collation processing, and the deviation amount (ΔH, Since the collation fingerprint image Gs having ΔV) is updated as the registered fingerprint image Go, the registration fingerprint renewal along with the registration renewal in the case where the manner of placing the finger accompanying the fingerprint collation of the registered principal has changed habitually changes. Depending on the registration Also performed, kill is possible to prevent the possibility of causing erroneous determination and collation processing time delay, such as false rejection in fingerprint collation processes after this.

  In the fingerprint collation / update function of the first to third embodiments, in any of the embodiments, the amount of horizontal translational deviation between the registered fingerprint image Go and the collated fingerprint image Gs determined to collate. Although it is configured to determine the presence or absence of registration update based on two deviation amounts of ΔH and vertical translation deviation amount ΔV or its frequency distribution [ΔH] [ΔV], fingerprint collation / update of the following fourth embodiment As will be described in terms of the function, three shift amounts of the horizontal translation shift amount ΔH and the rotation shift amount ΔR added to the vertical translation shift amount ΔV or the frequency distribution [ΔH] [ΔV] of the collation matching fingerprint image Gs. Based on [ΔR], it may be configured to determine whether or not there is a registration update.

(Fourth embodiment)
FIG. 9 is a flowchart showing fingerprint collation / update processing of the fourth embodiment in the fingerprint image collation apparatus.

  In the fingerprint collation / update process of the fourth embodiment, the same processing steps as those of the fingerprint collation / update process of the second embodiment in FIG. 7 are denoted by the same step symbols and description thereof is omitted.

  That is, in the fingerprint collation / update process according to the fourth embodiment, when it is determined that the person matches according to the fingerprint collation process (step AS) (step A1), the horizontal and vertical directions of the collation fingerprint image Gs with respect to the registered fingerprint image Go In addition to the translational deviation amounts (ΔH, ΔV), the rotational deviation amount (ΔR) is calculated and stored in the horizontal direction deviation amount memory 13g, the vertical direction deviation amount memory 13h, and the rotational direction deviation amount memory 13k (step A2 ′). .

  Further, frequency distributions [ΔH, ΔV, ΔR] of the horizontal and vertical translational deviation amounts (ΔH, ΔV) and rotational direction deviation amounts (ΔR) are calculated, and the horizontal deviation frequency distribution memory 13i and the vertical deviation frequency distribution are calculated. It is stored in the memory 13j and the rotational deviation frequency distribution memory 13l (step A2a ').

  Then, the horizontal translational deviation amount ΔH, the vertical translational deviation amount ΔV, and the rotational direction deviation amount ΔR of the collated fingerprint image Gs that has been determined to collate with the registered fingerprint image Go are allowed threshold values Th−H, TH−. If it is determined that it is less than V, TH-R, this collation fingerprint image data Gs is overwritten and updated in the registered fingerprint image memory 13b as registered fingerprint image data Go (step A3 '→ A4).

  On the other hand, it is determined that any one of the horizontal translational deviation amount ΔH, the vertical translational deviation amount ΔV, and the rotational direction deviation amount ΔR is not less than the allowable threshold values Th-H, TH-V, and TH-R. If the frequency distribution [ΔH] or [ΔV] or [?] Corresponds to a large deviation amount (ΔH or ΔV or ΔR) that is not less than the allowable threshold Th-H or TH-V or TH-R, It is determined whether or not [Delta] R] is equal to or higher than the frequency threshold Th-F indicating that it is habitual (step A3 '→ A5').

  Then, any one of the horizontal and vertical translational deviations (ΔH, ΔV) and the rotational deviation (ΔR) of the verification fingerprint image Gs is less than the allowable thresholds Th-H, TH-V, TH-R. Even when it is determined that it does not fit, the frequency distribution [ΔH] or [ΔV] or [ΔR] corresponding to the large deviation amount (ΔH or ΔV or ΔR) indicates that it is habitual. If it is determined that the threshold Th-F is exceeded, it is predicted that a collation fingerprint image Gs having the same horizontal deviation ΔH, vertical deviation ΔV, and rotation deviation ΔR will be obtained in the future collation processing. The verification fingerprint image Gs having a large deviation amount (ΔH) (ΔV) (ΔR) that is equal to or greater than the frequency threshold Th-F is overwritten and stored in the registered fingerprint image memory 13b as a registered fingerprint image Go (step A5). → A6).

  Then, the horizontal deviation frequency distribution data, the vertical deviation frequency distribution data, and the rotation deviation frequency distribution data stored in the horizontal deviation frequency distribution memory 13i, the vertical deviation frequency distribution memory 13j, and the rotation deviation frequency distribution memory 13l are cleared. And is initialized (step A7 ').

  On the other hand, any one of the horizontal and vertical translational deviations (ΔH, ΔV) and the rotational deviation (ΔR) of the verification fingerprint image Gs is less than the allowable thresholds Th-H, TH-V, TH-R. The frequency indicating that the frequency distribution [ΔH] or [ΔV] or [ΔR] corresponding to the large amount of deviation (ΔH or ΔV or ΔR) is habitual when it is determined that it does not fit If it is determined that the threshold value Th-F is not exceeded, the verification fingerprint image Gs confirmed with the registered person's identity is not overwritten and updated on the registered fingerprint image Go. The fingerprint collation / update process is terminated (step A5 ′ → end).

  Therefore, according to the fingerprint collation / update function of the fourth embodiment in the fingerprint image collation apparatus having the above-described configuration, when a collation match determination between the registered fingerprint image Go and the collation fingerprint image Gs is obtained, If the horizontal translational deviation amount ΔH, the vertical translational deviation amount ΔV, and the rotational direction deviation amount ΔR according to the placement are all less than the permissible threshold values Th-H, TH-V, TH-R, they are the same. The determined collation fingerprint image Gs is updated as a registered fingerprint image Go and corresponds to the large deviation amount ΔH or ΔV or ΔR even when it is not less than the allowable threshold values Th-H, TH-V, and TH-R. However, if the frequency distribution [ΔH] or [ΔV] or [ΔR] is equal to or higher than the frequency threshold Th-F indicating that it is habitual, the same level of horizontal deviation ΔH or Vertical deviation ΔV or no rotation It is predicted that a collation fingerprint image Gs having ΔR is obtained, and the collation fingerprint image Gs having a large deviation amount ΔH or ΔV or ΔR that is equal to or greater than the frequency threshold Th-F is updated as the registered fingerprint image Go. As a result, the registered fingerprint image Go not only is updated as needed according to the time-dependent change of the registered person's fingerprint, but the registered fingerprint image Gs that rarely includes a shift in the rotation direction is also registered. The image Go is not updated, and the registration method is updated when the finger placement associated with the fingerprint verification of the registered person has changed habitually. It is possible to prevent the possibility of causing a time delay.

  In the fingerprint collation / update function of the fourth embodiment, the rotation deviation amount ΔR or its frequency distribution [ΔR] of the collation matching fingerprint image Gs added as a condition for determining whether or not registration is updated is the same as that in FIG. The same conditions can be added to both the fingerprint collation / update function of the first embodiment and the fingerprint collation / update function of the third embodiment in FIG.

  Note that each processing method by the fingerprint image matching device described in each of the embodiments, that is, the fingerprint matching / updating process of the first embodiment shown in the flowchart of FIG. 4, and the fingerprint matching / updating process shown in the flowchart of FIG. Fingerprint verification process according to the second embodiment, the fingerprint verification / update process of the second embodiment shown in the flowchart of FIG. 7, the fingerprint verification / update process of the third embodiment shown in the flowchart of FIG. 8, and the fourth embodiment shown in the flowchart of FIG. Each method such as fingerprint collation / update processing is a program that can be executed by a computer, such as a memory card (ROM card, RAM card, etc.), magnetic disk (floppy disk, hard disk, etc.), optical disc (CD-ROM). , DVD, etc.), an external recording medium 18a such as a semiconductor memory and the like can be distributed. Various computer terminals having a fingerprint reading function read the program stored in the external recording medium 18a into the storage device 12 by the storage medium reading unit 18, and the operation is controlled by the read program. The fingerprint collation / update function described in each of the above embodiments can be realized, and the same processing can be executed by the method described above.

  The program data for realizing each of the above methods can be transmitted on the communication network (Internet) 20 in the form of a program code, and a computer terminal (program server) connected to the communication network (Internet) 20 It is also possible to implement the above-described fingerprint collation / update function by taking in the program data from the storage device 22 of 21).

  Note that the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the invention at the stage of implementation. Further, each of the embodiments includes inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in each embodiment or some constituent features are combined, the problems described in the column of the problem to be solved by the invention can be solved. When the effects described in the column of the effect of the invention can be obtained, a configuration in which these constituent elements are deleted or combined can be extracted as an invention.

The block diagram which shows the structure of the electronic circuit of the fingerprint image collation apparatus which concerns on embodiment of the image collation apparatus of this invention. The figure which shows the data memory ensured in RAM13 of the said fingerprint image collation apparatus. The frequency distribution (Histogram) data of each of the horizontal translational deviation (ΔH) and the vertical translational deviation (ΔV) stored in the horizontal deviation frequency distribution memory 13i and the vertical deviation frequency distribution memory 13j of the fingerprint image collating apparatus are shown. FIG. 6 is a flowchart showing fingerprint collation / update processing of the first embodiment in the fingerprint image collation apparatus. 7 is a flowchart separately showing fingerprint collation processing accompanying fingerprint collation / update processing of the fingerprint image collation apparatus. The arrangement state of the reference mask areas (templates) t1 and t2 on the registered fingerprint image Go and the maximum correlation positions (areas) k1 and k2 on the verification fingerprint image Gs are shown. Figure. 9 is a flowchart showing fingerprint collation / update processing of the second embodiment in the fingerprint image collation apparatus. 10 is a flowchart showing fingerprint collation / update processing of the third embodiment in the fingerprint image collation apparatus. 14 is a flowchart showing fingerprint collation / update processing of the fourth embodiment in the fingerprint image collation apparatus.

Explanation of symbols

11 ... CPU
12 ... Storage device 13 ... RAM
13a ... Read image (G) memory 13b ... Registered fingerprint image (Go) memory 13c ... Verification fingerprint image (Gs) memory 13d ... Reference mask size / position information (t1, t2, ..., tn) memory 13e ... Maximum correlation position information (K1, k2,..., Kn) Memory 13f ... Verification determination result memory 13g ... Horizontal displacement (ΔH) memory 13h ... Vertical displacement (ΔV) memory 13i ... Horizontal displacement frequency distribution [ΔH] memory 13j ... Vertical displacement Frequency distribution [ΔV] memory 13k Rotational direction deviation amount (ΔR) memory 13l Rotational deviation frequency distribution [ΔR] memory 14 ... Image reading device 15 ... Display unit 16 ... Input unit 17 ... Bus 18 ... Storage medium reading unit 18a ... External storage medium 19 ... Transmission control unit 20 ... Communication network (Internet)
21 ... Program server 22 ... Server storage device Go ... Registered fingerprint image data Gs ... Verification fingerprint image data Th-H ... Horizontal deviation tolerance threshold Th-V ... Vertical deviation tolerance threshold Th-R ... Rotation deviation tolerance threshold Th-F ... Custom Frequency threshold for misalignment judgment

Claims (6)

Image collation means for collating a registered image and a collation image to determine whether or not both match,
A deviation amount calculating means for calculating a positional deviation amount of the matching image with respect to the registered image when the image matching means determines that the registered image matches the matching image;
When the positional deviation amount of the collation image with respect to the registered image calculated by the deviation amount calculation unit is less than a preset positional deviation threshold, the collation image is updated as a new registered image that replaces the previous registered image. Registration update means to
Every time the registered image and the collated image are determined to be coincident by the image collating unit, a deviation frequency for storing the positional deviation amount of the collated image with respect to the registered image calculated by the deviation amount calculating unit as a distribution of positional deviation frequencies. Distribution storage means;
Even when the positional deviation amount of the collation image with respect to the registered image calculated by the deviation amount calculating means is greater than or equal to a preset positional deviation threshold, the deviation frequency distribution storage means corresponding to the positional deviation amount is stored. When the position deviation frequency is equal to or higher than a preset frequency threshold, the high-fluctuation frequency registration update unit updates the collation image as a new registered image that replaces the previous registered image;
An image collating apparatus comprising: The image collation apparatus according to claim 1 , wherein the positional deviation amount is a translational positional deviation amount of a collation image with respect to a registered image. The image collation apparatus according to claim 1 , wherein the misregistration amount is a misregistration amount of rotation of the collation image with respect to the registered image. The positional deviation amount is a translational and rotational positional deviation amount of the collation image with respect to the registered image,
The registration update unit, when positional deviation amounts of both the translation and rotation of the match image relative to said registered image calculated by said deviation amount calculating means is less than a preset positional deviation threshold, the collation image it update as a new registration image to replace the registered image of up to,
Even when the amount of misregistration of both the translation and rotation of the collation image with respect to the registered image calculated by the misregistration amount calculation unit is equal to or greater than a preset misregistration threshold, when at least one of the displacement frequency of the stored translation and rotation becomes a preset frequency threshold or by the deviation frequency distribution storage means where corresponding to the displacement amount of the collation image until it image collating apparatus according to claim 1, characterized in that updating as a new registered image to replace the registered image. An image matching processing program for controlling a computer of an image matching device,
The computer,
Image collation means for collating a registered image and a collation image to determine whether or not they match,
A deviation amount calculating means for calculating a positional deviation amount of the verification image with respect to the registered image when the image verification means determines that the registered image and the verification image match;
When the positional deviation amount of the collation image with respect to the registered image calculated by the deviation amount calculation unit is less than a preset positional deviation threshold, the collation image is updated as a new registered image that replaces the previous registered image. Registration update means,
Each time the registered image and the collated image are determined to be coincident by the image collating unit, a deviation frequency for storing a positional deviation amount of the collated image with respect to the registered image calculated by the deviation amount calculating unit as a positional deviation frequency distribution. Distribution storage means,
Even when the positional deviation amount of the collation image with respect to the registered image calculated by the deviation amount calculating means is greater than or equal to a preset positional deviation threshold, the deviation frequency distribution storage means corresponding to the positional deviation amount is stored. A registration error update unit at a high deviation frequency that updates the collation image as a new registered image that replaces the previous registered image when the positional deviation frequency is equal to or higher than a preset frequency threshold;
A computer-readable image matching processing program designed to function as a computer. An image collation step for collating the registered image and the collation image to determine whether or not both match,
A deviation amount calculating step for calculating a positional deviation amount of the collation image with respect to the registered image when it is determined in the image collation step that the registered image and the collation image match;
When the positional deviation amount of the collation image with respect to the registered image calculated in the deviation amount calculating step is less than a preset positional deviation threshold, the collation image is used as a new registered image that replaces the previous registration image. A registration update step to update;
Each time the registered image and the collated image are determined to coincide with each other in the image collating step, a deviation frequency for storing the positional deviation amount of the collated image with respect to the registered image calculated in the deviation amount calculating step as a positional deviation frequency distribution. A distribution memory step;
Even when the positional deviation amount of the collation image with respect to the registered image calculated in the deviation amount calculating step is equal to or larger than a preset positional deviation threshold, the deviation frequency distribution storing step corresponding to the positional deviation amount stores the amount. When the position deviation frequency is equal to or higher than a preset frequency threshold, the registration update step at the time of high deviation frequency for updating the collation image as a new registered image replacing the previous registered image;
An image collating method characterized by comprising:

Images