JPH10187983A - Image collating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To surely collate images with simple configuration by dividing a binarized image obtained by binarizing a video signal into squares to form plural segments and executing various kinds of processing with the segments as a unit. SOLUTION: The plural segments are formed by dividing the image outputted from a thinning circuit 20 horizontally and vertically at equal intervals for the unit of eight pixels. A reference voltage generating circuit 15 is composed of a latch circuit for holding the reference data of respective segments and a digital/analog(D/A) converting circuit for performing D/A converting processing to the reference data and outputting them and outputs a reference voltage REF corresponding to the respective segments. Then, at a comparator circuit 14, the reference data for light quantity correction set to the reference voltage generating circuit 15 are corrected by correction data stored in a threshold value correction memory 24, the reference voltage REF is corrected in segment unit in accordance with the image-picked up result, and even when the surface state or the like of isosceles triangle prism 11 is changed, a binarized signal S1 is correctly outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image collating device and can be applied to, for example, a fingerprint collating device. The present invention provides a simple configuration by dividing a binarized image obtained by binarizing a video signal into meshes to form a plurality of segments, and performing various processes in units of the segments. So that image matching can be performed reliably.

[0002]

2. Description of the Related Art Conventionally, in a fingerprint collating apparatus comprising this kind of image collating apparatus, a fingerprint image obtained through an imaging means is used to extract a characteristic portion such as a fingerprint branch, a dot, a cut, etc. By extracting feature points, fingerprints are collated based on these feature points.

That is, in a conventional fingerprint collation device,
For example, a database of fingerprints to be collated in advance is created based on the feature points of each fingerprint and its coordinate values. On the other hand, a fingerprint image to be inspected is subjected to image processing to extract feature points. Further, the database is accessed by using the extracted feature points, and the fingerprint is collated based on the presence or absence of the corresponding feature points.

[0004]

However, in the conventional fingerprint collation, there is a problem that processing takes time. If fingerprint collation can be reliably performed in a short time, usability of this type of fingerprint collation device can be improved and it is considered to be convenient.

[0005] The present invention has been made in view of the above points, and an object of the present invention is to propose an image matching apparatus capable of reliably performing image matching with a simple configuration.

[0006]

According to the present invention, a binary image is divided into meshes to form a plurality of segments, and a threshold is set for each segment. To generate a binary signal.

In addition, the signal level of the binarized signal is detected for each segment, and an effective area obtained by imaging the object to be verified is detected from the binarized image.

Further, the signal level of the binarized signal in a predetermined segment of the plurality of segments is detected to detect the presence / absence of an object to be verified in the binarized image.

If the binarized image is divided into meshes to form a plurality of segments, and a threshold is set for each segment to generate a binarized signal, the binarized image is applied to, for example, fingerprint collation. In addition, it is possible to compensate for partial deformation of the fingerprint and the like, and it is possible to prevent deterioration of the matching accuracy due to the partial deformation.

If the signal level of the binarized signal is detected for each segment and an effective area obtained by imaging the object to be verified is detected from the binarized image, the effective area in units of this segment is determined. Various processes can be executed with reference to simplify the processes.

More specifically, unnecessary processing is performed by comparing the binarized image and the reference image only within the effective area, and extracting and judging a plurality of linear images from within the effective area. Can be omitted. Further, based on the coordinate data of the effective area, the inclination of the collation target is detected, and the position of the collation target,
By detecting the size of the effective area, these detection results can be used for collation to improve collation accuracy.

For segments other than the effective area, by setting the signal level of the binarized signal to a constant value, unnecessary areas can be masked and processed. Further, for each segment of the effective region, the signal level of the binarized signal is detected to detect a segment in which it is difficult to secure the ratio between the light portion and the dark portion to a predetermined value or more, thereby reducing the region that degrades the matching accuracy. It can detect and take various actions.

Further, if the signal level of the binarized signal in a predetermined segment of the plurality of segments is detected and the presence or absence of a collation target in the binarized image is detected, the detection result is used as a trigger, for example, to perform collation. Processing can begin.

[0014]

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

(1) Configuration of Example (1─1) Overall Configuration FIG. 1 is a block diagram showing the overall configuration of a fingerprint matching device according to an embodiment of the present invention. In the fingerprint collation device 1, the entire operation is controlled by the system control circuit 3 in response to a user operation input via the key input unit 2, so that a predetermined user can be set in advance.
The fingerprint data D1 is fetched via the fingerprint data input unit 4, and the fetched fingerprint data D1 is registered in a memory together with various user data to construct a fingerprint database 5.

On the other hand, when the user inputs a request for fingerprint collation, the fingerprint data D2 to be inspected is fetched via the fingerprint data input section 4, and the fetched fingerprint data D
2 is temporarily stored in the inspection fingerprint memory 6. Further, in response to the user's operation input via the key input unit 2, the fingerprint data D2 temporarily stored in the test fingerprint memory 6 and the corresponding fingerprint data D1 registered in the fingerprint database 5 are compared with the matching rate detection unit. Enter 7 Further, the degree of coincidence between the fingerprint data D1 and D2 is detected, and the system control circuit 3 determines whether the fingerprints coincide or not according to the degree of coincidence.

In this way, the fingerprint collation apparatus 1 is compiled into a database, and determines whether the fingerprint data D1 specified by the user ID input by the user matches the fingerprint data D2 input via the fingerprint data input unit 4. By determining whether or not the user who has requested the fingerprint collation is a corresponding person, a determination result is output. In addition, according to the degree of the coincidence, the fingerprint database 5 is updated if necessary, and the collation result is recorded. In the fingerprint collation device 1, a message is notified to the user via the display unit 8, and thereby a necessary instruction is notified to the user, for example, to input a user ID or the like.

(1-1) Fingerprint Data Input Unit FIG. 2 is a block diagram showing the fingerprint data input unit. The fingerprint data input unit 4 captures a fingerprint from a user's finger held at a predetermined finger placement position via a predetermined optical system, and obtains fingerprint data D1 and D2 based on the captured image.
Generate

Here, this optical system comprises an isosceles triangular prism 11 holding a finger for collecting a fingerprint on the bottom surface.
And a light source 12 for illuminating the bottom surface from the slope of the isosceles triangular prism 11 and a CCD camera 13 for imaging the illumination light of the light source 12 reflected on the bottom surface from the remaining slope. The light source 12 is configured by a light emitting diode array or the like.

In this optical system, the illumination light emitted from the light source 12 is totally reflected by the bottom surface of the portion where the skin is not in contact with the bottom surface of the isosceles triangular prism 11 and guided to the CCD camera 13. Conversely, for the portion where the skin is in contact with the bottom surface of the isosceles triangular prism 11, the illumination light is irregularly reflected at the bottom surface so that the illumination light does not enter the CCD camera 13. As a result, as shown in FIG. 3, the optical system outputs a bright and dark image corresponding to a fingerprint pattern from the CCD camera 13 in the form of a video signal SV.

In this embodiment, the CCD camera 13 captures an image of the finger with a 4: 3 aspect ratio so that the root direction from the tip of the finger is the horizontal direction of the image capturing result. In this case, since a blank portion around the finger is not attached to the bottom surface of the isosceles triangular prism 11, the image is captured at a high luminance level.

The comparison circuit 14 compares the video signal SV with 2
The value is converted into a binary signal S1. Latch circuit 18
Samples the binarized signal S1 at a timing corresponding to each pixel of the CCD camera 13 with reference to a clock output from a timing generator (TG) 19, and outputs 1-bit image data DV1. The thinning-out circuit 20 intermittently takes in the image data DV1 and thins out the image data DV1 at a constant rate and outputs the thinned-out image data. The serial / parallel conversion circuit (S / P) 21 converts the 2-bit image data DV1 into 8-byte units. And output intermittently, so that 1-bit × 8-byte image data DV
1 is converted into fingerprint data D1 or D2 of 8 bits × 1 byte and output to the data bus BUS.

(1-1-1) Light amount correction Here, the comparison circuit 14 generates the binary signal S1 based on the reference voltage REF output from the reference voltage generation circuit 15, and in this embodiment, The light amount unevenness of the optical system is corrected by the reference voltage REF.

That is, as shown in FIG. 4 with reference to one horizontal scanning period, when such an optical system has unevenness in light amount, the signal level of the video signal SV resulting from the imaging changes due to the unevenness in light amount. (FIG. 4 (A)), when the binarization is performed using the constant reference voltage SL, the binarized signal S1 is generated such that the period during which the signal level rises is generally shortened in a peripheral portion where the light amount decreases. (FIG. 4B). That is, it is difficult to correctly generate a binarized signal, and the matching accuracy of the fingerprint is also deteriorated accordingly.

Therefore, in this embodiment, by detecting the signal level of the video signal SV without capturing any fingerprint (FIG. 4 (C)), the light amount unevenness is detected by this optical system. Further, a reference voltage REF is set by offsetting the detected signal level by a predetermined value toward the 0 level side, and a binary signal is generated by the reference voltage REF (FIGS. 4D and 4E). Thereby, the fingerprint matching device 1
In, the light amount unevenness is corrected to improve the fingerprint collation accuracy. Further, based on the comparison result between the reference voltage REF and the imaging result in the standby state, a dirt on the finger placement position and deterioration of the light source 12 are detected and a warning is issued, thereby also improving the fingerprint collation accuracy. It has been made like that.

More specifically, in this embodiment, as shown in FIG. 5, the image output from the thinning circuit 20 is divided at equal intervals in units of eight pixels in the horizontal and vertical directions. Form a segment of The reference voltage generation circuit 15 includes a latch circuit that holds reference data of each segment, and a digital-to-analog conversion circuit that performs digital-to-analog conversion processing on the reference data of each latch circuit at a corresponding timing and outputs the data. Output the corresponding reference voltage REF.

At the time of shipment from the factory, the reference data detects the signal level of the video signal SV output from the CCD camera 13 and is registered in advance in the light quantity correction memory 16 based on the signal level detection result. Further, when the power supply is started, it is set in the reference voltage generation circuit 15 by the system control circuit 3. As a result, in the fingerprint collation apparatus 1, as shown in FIG. 6, for example, when the light amount of the video signal SV is insufficient in a part around the center in the X direction, the reference voltage REF is adjusted to correspond to the change in the light amount. Switch,
Even with an optical system having a simple configuration having light amount unevenness, it is possible to correctly binarize an imaging result corresponding to the unevenness of a fingerprint.

Thus, the light quantity correction memory 16 is constituted by a non-volatile memory which holds reference data for light quantity correction.

(1-1-2) Correction of Threshold By the way, when the finger is placed on the isosceles triangular prism 11 as described above, the contact area of the skin changes due to the pressing force of the finger on the bottom surface. Thereby, the video signal SV
In this case, the overall signal level changes due to this pressing force. That is, as shown in FIG. 7, when the pressing force is large, the overall signal level is reduced so that the range of the black level is increased (FIG. 7 (A)). Means that the overall signal level increases as the white level range increases (FIG. 7).
(B)). A similar phenomenon occurs depending on the degree of sweating. As a result, in the fingerprint matching device 1, the binary signal S1 changes according to the pressing force or the like, and the fingerprint matching result obtained from the binary signal S1 changes.

The comparison circuit 14 corrects the reference data for light quantity correction set in the reference voltage generation circuit 15 by the system control circuit 3 with the correction data stored in the threshold value correction memory 24 to obtain an image. According to the result, the reference voltage REF is corrected for each segment, so that the binary signal S1 can be correctly output even if the pressing force, the surface state of the isosceles triangular prism 11, or the like changes.

Here, the correction data stored in the threshold value correction memory 24 selectively inputs the fingerprint data D1 output to the data bus BUS to the system control circuit 3,
In the system control circuit 3, the fingerprint data D1
The period T1 in which the signal level of the binarized signal S1 falls by detecting the logical level of
The value is set based on the measurement result of the fingerprint data D1 so as to fall within the range of 70% (FIG. 7C).

In the comparison circuit 14, the binary signal S1 is generated by the reference voltage REF corrected by the correction data.
And the contents of the threshold value correction memory 24 are further updated in the binarized signal S1. This series of processing is repeated, and when the sum of the periods T1 and T2 falls within a certain range in each segment. Then, the update processing of the threshold value correction memory 24 ends.

Thus, in the fingerprint collation apparatus 1, the signal level of the reference voltage REF, which is a threshold value, is set to an optimum value for each segment, and a correct collation result can be output even if the pressing force of the finger changes. It has been made like that.

(1-1-3) Switching of Threshold In the fingerprint collation device 1, the reference voltage R
When the EF is set, the fetching of the fingerprint data D1 and D2 into the fingerprint database 5 or the test fingerprint memory 6 starts, and the comparison circuit 14 determines the reference data in the light quantity correction memory 16 as shown in FIG. Reference voltage RE
After outputting the binarized signal S1 by F (FIGS. 8A and 8B1), the first and second references are displaced by a predetermined voltage to the positive side and the negative side with respect to this reference voltage REF. Voltage R
The binarized signal S1 is sequentially output by EF1 and REF2 (FIGS. 8A, 8B2 and 8B3). FIG. 8 shows a case where the light amount unevenness is not corrected by the reference voltage REF.

Here, the switching of the reference voltage REF is performed as follows.
By the system control circuit 3, the reference data in the light quantity correction memory 16 is further corrected by a predetermined value while being corrected by the contents of the threshold value correction memory 24, and set again in the reference voltage generation circuit 15. Be executed. As a result, in the fingerprint matching device 1, it is possible to effectively avoid deterioration of the matching accuracy due to the pressing force of the finger, a change in the surface state of the isosceles triangular prism 11, and the like, and to output a correct matching result.

(1-1-4) Output of Fingerprint Data When the fingerprint data D1 and D2 are output in this manner, the timing generator 19 uses the clock signal SV of the fingerprint collation device 1 based on the video signal SV. Is generated, and the cycle of the clock output to the thinning circuit 20 is switched according to an instruction from the system control circuit 3. As a result, in the thinning circuit 20, the fingerprint data D1, D2
The thinning rate is switched in accordance with the size of the finger to be collected. In the fingerprint matching device 1, when the finger is small, the thinning rate is reduced from the image data DV1 to reduce the fingerprint data D1 and D2. It is generated so that fingerprint collation can be performed reliably.

(1-1-5) Pulse Wave Detector The pulse wave detector 22 sends a pressure sensor 23 arranged on the side of the finger placement position and a detection result of the pressure sensor 23 to the system control circuit 3. And an output peripheral circuit.
Here, the pressure sensor 23 is configured to press the side of the finger together with a pressing mechanism (not shown) according to a command from the system control circuit 3 when the finger is placed at the finger placement position. A pressure detection result that changes according to the flow is output.

As described above, the pressure detection result changes in accordance with the pulsation of the blood flow, and the system control circuit 3 uses the pressure detection result (that is, the detection result of the pulse wave) to detect the finger. Biological reactions can be detected. Accordingly, the fingerprint matching device 1 executes the fingerprint registration process and the fingerprint matching process only when it can be determined that a human finger has been placed based on the pressure detection result, thereby improving security. Has been made. In this embodiment, the logic level of the fingerprint data D1 and D2 is monitored by the system control circuit 3, and a change in the logic level is used as a trigger to detect the placement of the finger. Only when it can be determined that a human finger has been placed on the basis of the detection result of step 22, a series of collation processing or the like is started.

(1-2) Fingerprint Database 5 and Test Fingerprint Memory 6 (1-2-1) Test Fingerprint Memory 6 FIG. 9 is a block diagram showing the fingerprint database 5 and test fingerprint memory 6 together with peripheral circuits. Here, the test fingerprint memory 6 holds the fingerprint data D1 and D2 output from the fingerprint data input unit 4 at the time of registering the fingerprint data D1 in the fingerprint database 5 and at the time of fingerprint collation.

In other words, the fingerprint data D1 and D2 input from the fingerprint data input unit 4 at the normal thinning rate by the address control executed by the memory control circuit 30 under the control of the system control circuit 3 and the test fingerprint memory 6. Stores fingerprint data D1 and D2 of 256 pixels in the horizontal direction and 128 pixels in the vertical direction, which are obtained by thinning out 1/2 of the pixels of the CCD camera 13.
On the other hand, when the thinning rate is switched and input according to the size of the finger, the input is selectively performed only for a partial area of the image formed by the fingerprint data D1 and D2.
As a result, fingerprint data D1, D of 256 pixels in the horizontal direction and 128 pixels in the vertical direction, as in the normal case, are obtained.
2 is stored. Thereby, in the fingerprint matching device 1,
An image captured with a variable thinning rate can be subjected to fingerprint collation by the same processing as usual.

Further, at this time, the test fingerprint memory 6 stores the fingerprint data D1 and D2 output from the fingerprint data input unit 4 in the erect image memory 6A, and sequentially switches the data bus BUS by the address control by the memory control circuit 30. The fingerprint data fetched into the erect image memory 6A via the erect image memory 6A is output to the image rotation circuit 31. Test fingerprint memory 6
Is an image rotation circuit 31 as shown in FIGS.
Output image of the original fingerprint data,
The fingerprint data D1 of the image rotated by 90 degrees is stored in the rotated image memory 6B. Thus, the fingerprint collation apparatus 1 can detect the fingerprint collation result by using the image obtained by rotating the original image by 90 degrees and performing the same processing as that of the original image.

That is, as shown in FIG. 12, the image rotation circuit 31 includes flip-flop circuits (FF) 31AA to 31HH for 8 × 8 bits connected in a matrix.
And a data bus BUS
Each of the bit data of the fingerprint data D1 and D2 input from these flip-flop circuits (FF) 31AA to 31
After sequentially transferring 8 bytes by HH, by outputting these bits in parallel, the arrangement of the fingerprint data D1 and D2 is changed by 90 degrees in units of segments. As a result, as shown in FIG. 13, the image rotation circuit 31 inputs the fingerprint data D1 and D2 in units of segments indicated by numerals 1 to 512, respectively, and then stores the fingerprint data D1 and D2 again in the storage means, thereby obtaining the image shown in FIG. Like 2 in 8 pixels
By processing the image data of the value, not only the arrangement direction of the segments but also the arrangement direction of the entire image can be changed.

In this way, the inspection fingerprint memory 6 holds the image of the fingerprint rotated by 90 degrees in this manner in the rotated image memory 6B with respect to the image stored in the erect image memory 6A. .

When the fingerprint data D1 is registered, the test fingerprint memory 6 inputs and holds a plurality of fingerprint data D1 for the same user, so that the fingerprint collation apparatus 1 uses the plurality of fingerprint data D1. Fingerprint data D1 which is most convenient for fingerprint collation is selectively registered in the fingerprint database 5 from among the data D1.

The test fingerprint memory 6 further includes a memory control circuit 30 controlled and executed by the system control circuit 3.
Address control by fingerprint control, fingerprint registration, fingerprint registration,
The fingerprint data D2 held in this way via the data bus BUS is selectively output in response to the operation of the matching ratio detection unit 7 described later. Thereby, in the fingerprint collation device 1, the fingerprint image is horizontally shifted in units of 8 pixels.
It can be cut out in the vertical and diagonal directions and output.

(1-2-2) Fingerprint Database 5 The fingerprint database 5 is composed of user data, fingerprint data D1, and the like registered in advance. That is, FIG.
As shown in FIG. 5, in the fingerprint database 5, a user ID is recorded for each registered user data, and fingerprint data D11, D12, and D13 of the index finger, middle finger, and little finger are registered for each user ID. It has been done. Thus, in the fingerprint collation device 1, even when the user has difficulty in collating the fingerprint of the middle finger due to, for example, injuries, or even when a matching result cannot be obtained with the middle finger,
Fingerprint data D2 input via fingerprint data input unit 4
The fingerprint data D11, D12, and D13 are compared with each other to confirm the identity of the user, thereby improving the usability.

At this time, the fingerprint data D11, D1
2. In D13, priority data is recorded, and fingerprint collation is sequentially performed in accordance with the priority data. As a result, in the fingerprint collation device 1, the waiting time until the collation result is obtained as a whole can be shortened, and the usability can be improved.

Further, the fingerprint database 5 records the thinning rate set by the thinning circuit 20 (FIG. 2) at the time of fingerprint registration as data of the magnification of each of the fingerprint data D11, D12, and D13. When the fingerprint data D2 is input, it can be captured at a corresponding magnification. The matching rate detected by the fingerprint matching device 1 is determined by the user I
The collation conditions are changed as necessary by recording the collation rate, and the fingerprint data D is recorded.
Re-registration of 11 to D13 can be prompted. In this way, the fingerprint collation device can reliably identify the person.

Thus, the fingerprint database 5 is stored in the memory control circuit 30 controlled by the system control circuit 3.
As in the case of the test fingerprint memory 6, the stored fingerprint data D11, D12, D13 and the like are selectively output and the contents are updated by the address control described above.

The rotating image memory for database 5B stores the fingerprint data D11 held in the fingerprint database 5,
An image rotation circuit 31 is applied to the images based on D12 and D13.
Holds a fingerprint image rotated by 90 degrees. Thus, in the fingerprint collating apparatus 1, the fingerprint image rotated 90 degrees is stored in the rotation image memory 6B and the database rotation image memory 5 corresponding to the erect fingerprint image held in the fingerprint database 5 and the inspection fingerprint memory 6, respectively.
B, and the image rotated by 90 degrees can be processed in the same manner as the original erect image.

(1-3) Collation Rate Detector 7 FIG. 16 is a block diagram showing the collation rate detector 7. The collation rate detection unit 7 has eight collation units 40A to 40H, and partially sets the fingerprint data D2 to be inspected in each of the collation units 40A to 40H. Further, the matching rate detection unit 7
Supplies the fingerprint data D1 to each of the collating units 40A to 40H sequentially so that the image based on the set fingerprint data is two-dimensionally scanned on the image based on the fingerprint data D1.
At each scanning position, the degree of correlation between the fingerprint data D2 and the fingerprint data D1 is detected.

Further, based on the detection result, coordinate values having a high correlation are sequentially registered in the coordinate group memory 49, and a collation rate is calculated in the system control circuit 3 from the positional relationship of the registered coordinate groups. A match is determined. In addition, since each of the collating units 40A to 40H has the same configuration, in FIG.
Are described in detail, and the other collating units 40B to 40H
Is omitted from the description. In addition, at the time of registration, the collation rate detection unit 7 detects a correlation value having a collation rate between a plurality of images based on the fingerprint data D1 instead of the determination between the fingerprint data D1 and D2.

That is, in each of the collating units 40B to 40H, the fingerprint data D2 stored in the test fingerprint memory 6 is set in the latch circuit 41 for each 8 bytes. Here, as shown in FIG. 17, the fingerprint data D2 set in each of the collating units 40B to 40H is converted from the image based on the fingerprint data D2 by the address control of the memory control circuit 30 (FIG. 9) under the control of the system control circuit 3. , A linear image is cut out, and fingerprint data D2 based on the linear image is allocated. In other words, on this original image,
Fingerprint data D2 that is continuous in the vertical or oblique direction is assigned. Further, in each of the collating units 40B to 40H,
As described above, fingerprint data D2 whose logic level switches in the continuous direction is assigned a predetermined number of times or more by the prior inspection by the system control circuit 3 so that the continuous directions become parallel.

The fingerprint data D2 set in each of the collating units 40B to 40H in this manner is compared with the collating units 40A to 40H.
At 40H, common fingerprint data D1 is sequentially supplied. At this time, the fingerprint data D1 is stored in the system control circuit 3
According to the address control of the memory control circuit 30 (FIG. 9) by the control of (1), when the direction in which the pixels of the fingerprint data D2 are arranged in the horizontal direction corresponding to the fingerprint data D2 set in the collation units 40B to 40H, Continuous fingerprint data D1 is supplied in 8-bit units according to the raster scanning order.

In each of the collating units 40A to 40H, a parallel / serial conversion circuit (P / S) 42 converts the sequentially input fingerprint data D1 into serial data and outputs the serial data to the shift register 43 at a constant cycle. Shift register 43
Holds the serial data for 64 bits, sequentially shifts the bits in synchronization with the serial data output from the parallel-serial conversion circuit 42, and outputs the bit-shifted data in 64-bit parallel.

The latch circuit 41 outputs the latched 8-byte fingerprint data D2A to D2H in 64-bit parallel. The comparison circuit 44 is constituted by 64 exclusive OR circuits, and outputs a comparison result of each bit between the output data of the shift register 43 and the output data of the latch circuit 41.

Thus, each of the collating units 40A to 40H makes the fingerprint data D2 so that the pixel values continue in the horizontal direction, for example.
When A to D2H are set, as shown in FIG.
The linear image formed by the fingerprint data D2A to D2H is raster-scanned on the image formed by the fingerprint data D1, and the overlapping fingerprint data D1 and D2A to D2 are scanned.
Between 2Hs, it is determined whether or not each bit matches.

The counter 45 counts the comparison results of the 64 systems obtained by the comparison circuit 44 in this way, and outputs a count value. Thus, the counter 45 detects the degree of correlation between the linear image formed by the fingerprint data D2A to D2H and the image formed by the fingerprint data D1 overlapping the linear image. If the overlapping images match exactly,
A count value of 64 is output.

In the register 46, reference value data for determination is set by the system control circuit 3, and a comparison circuit 47 is set.
Compares the reference value data set in the register 46 with the count value of the counter 45, and performs raster scanning in this way to overlap the overlapped fingerprint data D1 and D2A to D2H.
Among them, judgment data D4A to D4H in which the signal level rises when the degree of correlation is large is output.

The coordinate group memory 49 stores judgment data D4A to
When D4H rises, the address data AD output from the memory control circuit 30 is corrected and recorded, so that the relative position information between the fingerprint data D2A to D2H and D1 at the timing when each of the determination data D4A to D4H rises is obtained. It is recorded for each of the collating units 40A to 40B.

Thus, the coordinate group memory 49 raster scans the linear image on the image of the fingerprint data D1, corrects the address data AD at the timing when the determination data rises, and takes in the linear image. Then, a position indicating a high degree of coincidence with the image of the overlapping fingerprint data D1 is recorded by the coordinate value of the fingerprint data D1.

In this way, for an image having a repetitive shape such as a fingerprint, scanning the line-cut image in this way to determine the degree of correlation between the image and the overlapping image reveals that a high degree of coincidence is obtained at many places. You can get a degree.
Therefore, in the coordinate group memory 49, each of the matching units 40A to 40H is stored.
Each time, a plurality of coordinate data are recorded. However, if the two images are fingerprints of different persons, the coordinate group memory 49 stores coordinate values that are irrelevant between the matching units 40A to 40H, whereas the two images are fingerprints of the same person. In this case, the relative coordinate values held in the relative positional relationship between the fingerprint data D2A to D2H described above with reference to FIG. Thus, in this embodiment, data of the collation rate is generated from the coordinate group registered in the coordinate group memory 49 using this relationship.

Further, in this embodiment, the coordinate group memory 49 divides the storage area and stores the coordinate data of the erect image and the coordinate data of the rotated image as shown in FIG. Here, the coordinate group memory 49 stores the fingerprint data D1 in the fingerprint database 5 and the erect image memory 6A.
And the coordinate data detected by D2 are recorded as the coordinate data of the erect image, and the fingerprint data D1 stored in the corresponding rotated image memory 5B and rotated image memory 6B for the database.
And the coordinate data detected by D2 is recorded as coordinate data of the rotated image.

Further, in this embodiment, the fingerprint collation apparatus 1 is configured to set the linear image to each of the collation units 40A to 40B while sequentially tilting them. The coordinate group memory 49 stores the erect image and the rotated image. In the coordinate data of
The coordinate data is recorded for each inclination angle. Thus, the fingerprint collation apparatus 1 comprehensively determines the coordinate data recorded in this way and collates the fingerprint, thereby improving the precision of fingerprint collation.

Further, in this embodiment, the coordinate group memory 49 is configured to hold a plurality of sets of such coordinate data of the erect image and coordinate data of the rotated image. The coordinate group memory 49 uses the plurality of systems as reference voltages REF, REF1, and RE in the fingerprint data input unit 4.
In accordance with the switching of F2 and the switching of the area (hereinafter referred to as a frame) from which the linear image is cut out, the data is held, and the coordinate data is recorded each time the reference voltage and the frame are switched. .

Thus, in the fingerprint collation apparatus 1, the coordinate data detected corresponding to the switching of the reference voltage and the switching of the frame are comprehensively determined to improve the collation result.

(1-4) System Control Circuit The system control circuit 3 is composed of a microcomputer, controls the entire operation of the fingerprint collating apparatus 1, registers the fingerprint data D1 in the fingerprint database 5, and registers the fingerprint data. The fingerprint collation processing is executed based on the database 5.

(1-4-1) Correction of Uneven Light Amount and Detection of Abnormality of Optical System FIG. 20 is a flowchart schematically showing the processing procedure of the system control circuit 3. The system control circuit 3
When the power is turned on, from step SP1 to step SP
The process proceeds to step S2, where the reference data registered in the light quantity correction memory 16 is set in the reference voltage generation circuit 15. Accordingly, the system control circuit 3 sets the reference voltage REF of the comparison circuit 14 so as to correct the light amount unevenness of the optical system, so that a correct collation result can be obtained even with an optical system having a simple configuration. .

Subsequently, the system control circuit 3 proceeds to step S
Moving to P3, an optical system operation check process is executed. Here, as shown in FIG. 21, in this optical system operation confirmation processing,
The system control circuit 3 proceeds from step SP4 to step S
The process proceeds to P5, where the output data of the serial / parallel conversion circuit 21 is fetched via the data bus BUS (FIG. 2), and the logical level is counted for each segment. Here, the counting of the logic level is executed by counting the number of falling bits of the logic level.

That is, when no finger is pressed on the bottom surface of the isosceles triangular prism 11 in the fingerprint data input unit 4, the illumination light of the light source 12 is totally reflected on the bottom surface and is imaged. The logic level of the output data is kept at the H level. On the other hand, when the bottom surface is dirty, the illumination light is diffusely reflected, so that the luminance level of the imaging result is partially reduced according to the degree of the contamination, and the logical level of the corresponding output data is changed to the L level. You will fall. When the characteristics of the light source 12 are deteriorated from the characteristics at the time of shipment from the factory and the light amount unevenness becomes severe,
It becomes difficult to perform correction depending on the reference data set in the light amount correction memory 16, and in this case, the logic level of the corresponding output data falls to the L level.

Thus, based on the count result, the system control circuit 3 determines in the following step SP6
It is determined whether or not this count value has exceeded a predetermined value. Here, if a positive result is obtained, the system control circuit 3 proceeds to step SP7, displays a maintenance call via the display unit 8, and then proceeds to step SP8 to end this processing procedure. Here, the maintenance call is executed by displaying a message urging the cleaning of the finger placing portion, and when the maintenance call is still displayed after the cleaning, a message notifying the maintenance worker is displayed.

As a result, in the fingerprint collation device 1, the light source 12
The fingerprint matching accuracy is prevented from deteriorating due to the characteristic deterioration of the above, dirt on the isosceles triangular prism 11, and the like, and the fingerprint matching accuracy is improved accordingly.

On the other hand, if a negative result is obtained in step SP6, the system control circuit 3 proceeds directly to step SP8 and ends this processing procedure.

When the optical system operation confirmation processing is executed as described above, the system control circuit 3 proceeds to step SP10 (FIG. 20) and determines whether a fingerprint registration request has been input. Here, in this embodiment, a fingerprint registration request can be input by inputting a predetermined command, a reciting number or the like via the key input unit 2, and the fingerprint data D1 of each user is input by the fingerprint registration request. Register in the fingerprint database 5. When this fingerprint registration request is input, the system control circuit 3 obtains a positive result in step SP10, moves to step SP11, executes fingerprint registration processing, and returns to step SP3.

On the other hand, if a fingerprint registration request has not been input, the system control circuit 3 proceeds from step SP10 to step SP12, and determines whether or not a fingerprint collation request has been input by operating the key input unit 2. Here, if a negative result is obtained, the system control circuit 3 proceeds to step S
Return to P3. On the other hand, when the system control circuit 3 detects a fingerprint collation request, the system control circuit 3
Moves from step SP12 to step SP13, executes fingerprint collation processing, and returns to step SP3.

(1-4-2) Fingerprint Registration Process FIG. 22 is a flowchart showing the procedure of this fingerprint registration process. In this fingerprint registration process, the system control circuit 3 proceeds from step SP20 to step SP21 and displays a message via the display unit 8 (FIG. 1).
The user is prompted to place the index finger at the finger placement position. Subsequently, the system control circuit 3 proceeds to step SP22, determines whether or not a finger has been placed at the finger placement position, and if a negative result is obtained here, repeats step SP22.

In step SP22, the system control circuit 3 inputs the output data of the serial / parallel conversion circuit 21 via the data bus BUS (FIG. 2), and counts the logical level of the output data in segment units. Thus, whether or not a finger is placed at the finger placement position is detected based on the output data. This counting process is executed by counting the number of bits of the logical L level for a predetermined plurality of segments corresponding to substantially the center portion of the finger placement position, and determining whether the count value has exceeded a predetermined value. As a result, the system control circuit 3 executes the fingerprint registration process by using the imaging result as a trigger, thereby simplifying the operation of the fingerprint matching device 1 and improving the usability.

When a positive result is obtained in step SP22, the system control circuit 3 proceeds to step S22.
The process proceeds to P23, and it is determined whether or not a biological reaction is detected by the pulse wave detector 22. Here, if a negative result is obtained, the system control circuit 3 proceeds to step SP24 to end this processing procedure, whereas if a positive result is obtained in step SP23, the system control circuit 3 proceeds to step SP25 to execute actual registration processing. Execute the actual registration process. Thus, the system control circuit 3 registers the fingerprint data only when a biological reaction of the finger is detected, thereby improving security.

When the actual registration processing is completed, the system control circuit 3 proceeds to step SP26, and determines whether fingerprint data has been registered up to the little finger. Here, as described above with reference to FIG. 15, the fingerprint collation device 1 registers the fingerprint data of the index finger, the middle finger, and the little finger for each user. If it is not registered, this step SP is performed unless a special operation by the user is detected.
When a negative result is obtained in 26, the process proceeds to step SP27, instructs placing of the next finger through the display unit 8, and then returns to step SP22.

On the other hand, if the registration is completed up to the little finger, a positive result is obtained in step SP26, so that the system control circuit 3 proceeds from step SP26 to step SP24, and ends this processing procedure.

(1-4-3) Real Registration Processing FIGS. 23 and 24 are flowcharts showing the processing procedure of the real registration processing in the system control circuit 3. In this actual registration process, the system control circuit 3 executes step S
The process moves from step P30 to step SP31, where an effective area detection process is executed. Here, as shown in FIG. 25, the system control circuit 3 executes steps SP32 to SP
At 33, the number of bits at the logical H level is counted for one segment, and the number of pixels at which the luminance level rises in this segment is counted.

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P34, where it is determined whether or not this count value is equal to or less than a predetermined value, thereby determining whether or not the area where the luminance level rises in the corresponding segment is equal to or less than a certain value. Here, in the optical system of the fingerprint collating apparatus, when no finger or the like is placed on the slope of the isosceles triangular prism 11, the illumination light is totally reflected and the luminance level of the corresponding pixel rises. For a segment whose luminance level has risen to a certain value or more, a blank segment where no finger is placed can be determined.

When the negative result is obtained in step SP34, the system control circuit 3 proceeds to step S34.
The process moves to P35, where this segment is set as a segment in the blank area, and then the process moves to step SP36. On the other hand, if a positive result is obtained in step SP34, the system control circuit 3 proceeds to step SP37, sets this segment in the effective area, and then sets in step SP36
Move on to The effective area in this case means an area where a finger is considered to be placed.

In step SP36, the system control circuit 3 determines whether or not a series of processes in the effective area detection process has been completed for all segments. If a negative result is obtained here, the process proceeds to step SP38.
Switching the processing target to the next segment and step SP3
Return to 3. On the other hand, when a series of processing is completed for all segments, the process moves from step SP36 to step SP39, and returns to the main routine.

Thus, as shown in FIG. 26, the system control circuit 3 sets an effective area usable for fingerprint collation in the imaging result in units of segments, and thereafter sets various areas based on this effective area as a reference. By executing the process of
These various processes can be simplified.
In FIG. 26, a segment marked with a symbol “x” is a segment of a blank area.

Subsequently, the system control circuit 3 proceeds to step S
Moving to P40 (FIG. 23), a finger position determination process is performed.
Here, as shown in FIG. 27, the system control circuit 3 proceeds from step SP41 to step SP42, and determines whether the number of segments in the effective area is equal to or less than a predetermined value. If a negative result is obtained here, in this case, the placement position of the finger is considered to be abnormal, and the system control circuit 3 proceeds to step SP43, where the display unit displays the finger again. After outputting the warning through the control unit 8, the process returns to the valid area detection processing.

On the other hand, if a positive result is obtained in step SP42, it can be determined that the placement position of the finger is correct in this case, so that the flow shifts to step SP44 and returns to the main routine. Thus, in the fingerprint matching device 1, the fingerprint data is registered only when the user puts the finger correctly, thereby improving the accuracy of the fingerprint matching.

Subsequently, the system control circuit 3 proceeds to step S
In P46 (FIG. 23), zoom processing is executed. As shown in FIG. 28, in this zoom processing, the system control circuit 3 executes steps SP47 to SP48.
Then, it is determined whether the number of segments in the effective area is equal to or smaller than a predetermined value. If the number of segments in the effective area is equal to or larger than the predetermined value, the finger is imaged in a size large enough for fingerprint collation, and the system control circuit 3 proceeds to step SP48.
Then, the routine goes to step SP49 and returns to the main routine.

On the other hand, if an affirmative result is obtained in step SP48, it is considered that the finger is imaged small, and the system control circuit 3 proceeds to step SP50 and performs thinning in the thinning circuit 20 (FIG. 2). It is determined whether the rate can be reduced. Here, if the thinning rate can be reduced, the system control circuit 3 proceeds to step SP51, switches the operation of the timing generator 19 to reduce the thinning rate, thereby enlarging the image based on the fingerprint data D1, and then enabling the step SP31. The process returns to the area detection processing.

Thus, the system control circuit 3 performs an effective area detection process on the enlarged fingerprint image again.
After executing the finger position inversion processing, this zoom processing is executed. Thus, in the fingerprint collation apparatus 1, the magnification of the optical system is switched as necessary, so that the fingerprint collation can be reliably performed.

If the finger is abnormally dry, the illumination light may not be irregularly reflected on the bottom surface of the isosceles triangular prism 11 (FIG. 2) even when the finger is pressed. In this case, even if the thinning rate is reduced in this way, a negative result may be obtained in step SP50, making it difficult to perform fingerprint matching correctly. Thereby, the system control circuit 3
Moves to step SP52 in this case, outputs a confirmation message to the user, and then moves to step SP53 to end a series of fingerprint registration processing. Thus, even in such a case, the fingerprint collation device 1 can surely perform fingerprint collation by re-capturing a fingerprint image.

Subsequently, the system control circuit 3 proceeds to step S
In P55 (FIG. 23), a threshold value correction process is performed. As shown in FIG. 29, in this threshold value correction processing, the system control circuit 3 proceeds from step SP56 to step SP57, and executes logic L for one segment.
Count the number of bits in the level. Here, in this embodiment, when the illumination light is irregularly reflected at the bottom surface of the isosceles triangular prism, the luminance level of the corresponding pixel falls, and the bit of the corresponding fingerprint data D1 becomes logical L.
By falling to the level, the system control circuit 3
Detects the area of the segment where the finger is in close contact with the bottom surface of the isosceles triangular prism by counting the number of bits.

Subsequently, the system control circuit 3 proceeds to step S
Proceeding to P58, it is determined whether or not this count value is included in the range of 30 to 70% for the number of pixels (8 × 8) of the segment. If a negative result is obtained here,
Since it is considered that the fingerprint image is locally distorted, the system control circuit 3 proceeds to step SP59,
After updating the corresponding correction data in the threshold value correction memory within the predetermined range, the process proceeds to step SP60. On the other hand, if a positive result is obtained in step SP58, the system control circuit 3 directly proceeds to step SP60.

In step SP60, the system control circuit 3 determines whether a series of processes in the threshold value correction process has been completed for all the segments set in the effective area, and a negative result is obtained here. When it is
Move to step SP61. Here, the system control circuit 3
Returns to step SP57 after switching the processing target to the next segment. On the other hand, when a series of processes in the threshold value correction process is completed for all segments, the system control circuit 3 proceeds to step SP6.
0 to obtain a positive result, the step S
The process moves to P62 and returns to the main routine. As a result, in the fingerprint collation apparatus 1, the contents of the threshold value correction memory 24 are set as described above with reference to FIG.

Subsequently, the system control circuit 3 proceeds to step S
In P65 (FIG. 23), blackout processing is executed.
That is, when the finger is abnormally wet due to sweat or the like, the illumination light is irregularly reflected in a certain area on the bottom surface of the isosceles triangular prism 11. In this case, even if the voltage of the reference voltage REF is varied within a certain range in the above-described threshold value correction processing, the luminance level of the corresponding pixel remains falling in the corresponding segment, and the black-out image Is obtained. Therefore, in this segment, it is difficult to correctly perform fingerprint collation.

As a result, the system control circuit 3
As shown in steps SP66 to SP67,
After counting the number of bits of the logical L level for one segment, the process proceeds to step SP68 to determine whether or not the count value is equal to or less than a predetermined value.

If a negative result is obtained here, the system control circuit 3 proceeds to step SP69, sets this segment to a blackout segment, and then proceeds to step SP7.
Move to 0. On the other hand, if a positive result is obtained in step SP68, the process directly proceeds to step SP70. As a result, the system control circuit 3 excludes the blackout segment from the processing target in the subsequent series of processing, thereby improving the fingerprint matching accuracy.

The system control circuit 3 determines in step SP70
, It is determined whether or not a series of processes in the blackout processing has been completed for all segments, and if a negative result is obtained here, the process proceeds to step SP71 to set the next segment of the effective area as a processing target. , Step S
It returns to P67. As a result, the system control circuit 3 detects the blackened segments for all the segments in the effective area, and then proceeds from step SP70 to step SP7.
Move to 2.

In this step SP72, the system control circuit 3 determines whether or not the number of blackout segments is equal to or less than a predetermined value. That is, if the number of blackout segments exceeds a certain value, an area required for fingerprint collation cannot be secured. In this case, the system control circuit 3 determines in step S
The process moves from P72 to step SP73 and outputs a warning to the user that the finger is to be cleaned and the finger is to be placed again, after which the process returns to step SP31.

On the other hand, if a positive result is obtained in step SP72, the system control circuit 3 proceeds to step SP74 and returns to the main routine. Accordingly, in the case where the finger is wet or the like, the fingerprint collation apparatus 1 prompts the user to perform a fingerprint registration operation again, captures the fingerprint image to be registered with high image quality, and improves the precision of fingerprint collation accordingly. Has been made.

Subsequently, the system control circuit 3 proceeds to step S
The process shifts to P80 (FIG. 23) to execute angle detection processing. That is, the user may place the finger at the finger placement position while tilting the finger. Therefore, the system control circuit 3
Corresponds to step SP in the angle detection process shown in FIG.
The process moves from step 81 to step SP82, and an edge is detected in the effective area, for example, in the Y direction, as shown in FIG. Further, the system control circuit 3 proceeds to step SP3, where, as shown in FIG. 33, a segment pair corresponding to the vertical direction is detected from the outermost segment, and the center coordinates of the segment pair are sequentially detected to thereby detect , The inclination θ about the center line of the fingerprint image is detected.

Subsequently, the system control circuit 3 proceeds to step S
Moving to P84, by determining whether or not the angle θ is equal to or less than a predetermined range, it is determined whether the inclination of the finger is equal to or less than a predetermined range.
If a negative result is obtained here, the process proceeds to step SP85, and outputs a warning to the user that the finger is correctly placed, and then returns to step SP31. On the other hand, when the inclination of the finger is equal to or smaller than the predetermined range, the system control circuit 3 proceeds to step SP86 and returns to the main routine.

As a result, the fingerprint collation apparatus 1 takes in the fingerprint data D1 while keeping the inclination within a certain range, thereby improving the accuracy of fingerprint collation.

When the prior processing is completed in this way,
The system control circuit 3 proceeds to step SP88 (FIG. 24). Here, the system control circuit 3 sets the value of the variable n for counting the number of images to be taken to a value of 1 and then controls the memory control circuit 30 to store the fingerprint data D1 for one screen in the test fingerprint memory 6 in step SP89. Store.

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P90, where it is determined whether or not the value of the variable n is equal to the value N set in advance. If a negative result is obtained here, the process proceeds to step SP91. Return to SP89. Accordingly, the system control circuit 3 determines in steps SP89-SP90-S
The processing procedure of P91-SP89 is repeated N times, and N images are stored in the inspection fingerprint memory 6.

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P92, in which it is determined whether or not this series of processes has been completed for the reference voltage REF2. In this case, a negative result is obtained, and the process proceeds to step SP93 to perform switching control of the reference voltage in the reference voltage generation circuit 15. After that, the process returns to step SP88.

Thus, the system control circuit 3 inspects N images using the reference voltage REF set by correcting the reference data stored in the light quantity correction memory 16 with the correction data in the threshold value correction memory 24. After being stored in the fingerprint memory 6, N images are stored in the inspection fingerprint memory 6 by the reference voltage REF1 obtained by offsetting the reference voltage REF to the positive side by a predetermined voltage, and the reference voltage REF is further lowered by the predetermined voltage to the negative side. Reference voltage REF2 offset to
, The N images are stored in the inspection fingerprint memory 6.

By simply offsetting the reference voltage REF and setting the reference voltages REF1 and REF2,
The system control circuit 3 performs the light amount unevenness correction processing (FIG. 2) in the same manner as in the case where N images are captured by the reference voltage REF.
0, step SP2) and N threshold images are input to the test fingerprint memory 6 in a state where the threshold correction processing in segment units (FIG. 29) is executed.

In this reference voltage switching process, the system control circuit 3 counts the logical level of the output data output from the serial / parallel conversion circuit 21 for all the effective areas except for the blackout areas, and thereby, in these areas, The white level area is 30 to 7
The reference voltage is switched so as to fall within the range of 0 [%]. According to the results of the experiments, it is found that the reference voltage is switched in this range to capture a plurality of fingerprint images, and the fingerprints are compared based on the captured results as in the present embodiment to improve the accuracy of fingerprint matching. Was completed.

When N images have been input for the reference voltage REF2 in this manner, the system control circuit 3
Moves to step SP93 when a positive result is obtained in step SP92, and executes registered data selection processing. Here, the registration data selection process is a process of determining which fingerprint data D1 is to be registered in the fingerprint database 5 for the 3 × N images thus captured. In step SP94, the system control circuit 3 registers the fingerprint data D1 selected by the registration data selection processing in the fingerprint database 5, and then proceeds to step SP95 to end this processing procedure.

Thus, the system control circuit 3 sequentially executes this series of processing for the index finger, the middle finger, and the little finger (FIG. 22), assigns priorities in the order of the index finger, the middle finger, and the little finger, and registers them in the fingerprint database 5. (FIG. 15). At this time, the system control circuit 3 also records the magnification (thinning rate in the thinning circuit 20) of the selected fingerprint data D1. Thus, the system control circuit 3 can reproduce the state at the time of actual fingerprint collation and enrollment for each fingerprint data D1 made into a database in this way, thereby also improving the accuracy of fingerprint collation. This makes it possible to reduce the time required for verification.

(1-4-4) Registration Data Selection Processing FIG. 34 is a flowchart showing the above-described registration data selection processing. The fingerprint data D1 to be registered in the fingerprint database 5 is selected from D1.

That is, the system control circuit 3 proceeds from step SP100 to step SP101 and sets variables n and m to values 1 and 2, respectively. Where the variable n
And m are variables that specify the fingerprint data D1. The system control circuit 3 sequentially detects a correlation value between the fingerprint data D1 specified by the variable n and the fingerprint data specified by the variable m. Further, the system control circuit 3 detects the correlation value having the largest value from the correlation values thus detected, and sets the fingerprint data D1 based on the corresponding variable n as the fingerprint data D1 to be registered.

That is, in the subsequent step SP102, the system control circuit 3 executes a correlation value detection process to detect a correlation value between the fingerprint data D1 based on the variable n and the fingerprint data D1 based on the variable m. Further, the system control circuit 3 proceeds to step SP103, where the inspection fingerprint memory 6
It is determined whether or not all the fingerprint data D1 fetched in the above are specified by the variable m. If a negative result is obtained here, the process proceeds to step SP104, and after the variable m is incremented, the process returns to step SP102.

As a result, the system control circuit 3 sets the variable m
Are sequentially switched to sequentially detect a correlation value of one fingerprint data D1 specified by the variable n with respect to another fingerprint data D1.

On the other hand, if an affirmative result is obtained in step SP103, the system control circuit 3 moves to step SP105, in which the fingerprint data D1 fetched into the test fingerprint memory 6 has been all specified by the variable n. Determine whether or not. If a negative result is obtained here, the system control circuit 3 proceeds to step SP106, where the variable n is incremented, the variable m is initialized to a value 1, and the process returns to step SP102.

Thus, the system control circuit 3 detects the correlation value of the fingerprint data D1 of the variable n for all combinations of the fingerprint data D1 fetched into the test fingerprint memory 6.

When the correlation value detection processing is executed for all combinations in this manner, the system control circuit 3
When a positive result is obtained in step SP105, the process proceeds to step SP107, in which the fingerprint data D1 with the variable n having the largest correlation value is selected as the fingerprint data to be registered, and then the process proceeds to step SP108 to end this processing procedure. I do.

Thus, the fingerprint matching device 1 determines the image quality of the fingerprint image based on the correlation value, and registers the fingerprint data D1 most convenient for fingerprint matching in the fingerprint database 5. That is, in the fingerprint collation apparatus 1, fingerprint collation is performed based on a collation rate, which will be described later, which is detected by the same method as the correlation value. As a result, the system control circuit 3 executes a fingerprint collation process with the remaining fingerprint images, assuming that the plurality of fingerprint images taken into the inspection fingerprint memory 6 are sequentially registered in the fingerprint database 5, In this way, a fingerprint image suitable for fingerprint collation is selected by a determination method based on actual fingerprint collation, and the fingerprint collation accuracy is improved.

FIG. 35 shows the fingerprint data D1 of the variable m and the fingerprint data D of the variable m by the system control circuit 3.
9 is a flowchart illustrating a correlation value detection process for No. 1; In this processing procedure, the system control circuit 3 proceeds from step SP110 to step SP11 and
And the fingerprint data D1 specified by m are transferred to the image rotation circuit 31 in segment units, and the fingerprint data output from the image rotation circuit 31 is stored in the rotation image memory 6B. Thereby, the system control circuit 3
In this step SP111, the image to be processed is rotated by 90 degrees (FIGS. 10 and 11).

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P112, where 8-byte continuous image data in the horizontal direction are sequentially and intermittently selected from the upper side of the image based on the fingerprint data D1 from the fingerprint data D1 specified by the variable m. Further, the selected fingerprint data is output to the matching units 40A to 40B, respectively.
B is stored in the latch circuit 41 (FIG. 17). As a result, the system control circuit 3 cuts out a linear image from the image based on the fingerprint data D1, and sets it in the matching units 40A to 40H.

At this time, the system control circuit 3 sets the fingerprint data D1 by executing a predetermined judgment procedure so as not to cut out the linear image for the black-out segment described above with reference to FIG. This effectively avoids a decrease in fingerprint collation accuracy. Further, an area (that is, a frame) from which this linear image is cut is set based on the effective area detected in advance.

Further, in advance, the number of times the logical level is switched is counted for the fingerprint data to be linearly cut out,
The part where the count value is equal to or less than the predetermined value is excluded from the target. As a result, the linear cut-out portion crosses the fingerprint for a predetermined number or more, so that the cut-out region sufficiently contains significant information for fingerprint matching, and this also improves the accuracy of fingerprint matching. Has been made to be able to improve.

By the way, the system control circuit 3 sets the position shifted from the tip end of the effective area formed by the tip of the finger toward the base of the finger by a predetermined segment, and in the vertical direction, the center of the effective area substantially Then, a frame for cutting out the linear image is set so as not to protrude outside the effective area.

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P113, where the fingerprint data D1 specified by the variable n is successively arranged in the order of raster scanning.
The fingerprint data is sequentially transferred to the matching units 40A to 40H. As a result, the system control circuit 3 sets each collation unit 40A
-40H, on the image specified by the variable n,
The linear image is scanned in the raster scanning order, and the number of bits that match between the linear image and the image with the variable n that overlaps the linear image is determined by each of the matching units 40A to 40H.
(FIG. 16).

Further, the system control circuit 3 supplies a coordinate value at which the count value rises above the threshold value set in the register 46, that is, a position at which two overlapping images are determined to be very similar in raster scanning order. The coordinates of the fingerprint data are recorded in the coordinate group memory 49.

At this time, the system control circuit 3 switches the operation of the comparison circuit 44 so as to output the comparison result from a certain logical level for the corresponding bit of the fingerprint data determined as the invalid area and the blackened area. Further, the threshold value set in the register 46 is switched accordingly, and the criterion in the subsequent comparison circuit 47 is changed. As a result, the system control circuit 3 masks such invalid areas and black-out areas, and processes these areas. The system control circuit 3 supplies fingerprint data so that the linear image scans the effective area, thereby omitting unnecessary processing for fingerprint registration and simplifying the entire processing.

The system control circuit 3 then proceeds to step S
The process moves to P114 to execute a matching ratio detection process. Here, as shown in FIG. 36, FIG. 37, and FIG. 38, in the matching rate detection process, the system control circuit 3
Then, the process goes to step SP116 to set variables I, J, K, and Q to a value of 1. Here, I, J, and K are variables that specify the coordinate values of the matching units 40A, 40B, and 40C that have been fetched into the coordinate group memory 49, and Q is a variable that is detected in the partial matching rate calculation process. This variable specifies a combination of coordinate values.

The system control circuit 3 then proceeds to step S
P117, where three coordinate values (XIA, YIA), (XJB, YJ) specified by variables I, J, K
B), (XKC, YKC), coordinate values (XIA, YIA) and coordinate values (XJB, YJB), coordinate values (XJB, YJB), coordinate values (XKC, YKC), coordinate values (XIA, YIA) And the coordinate values (XKC, YKC) are set, and among these three combinations, the number of combinations that satisfies the relative positional relationship of the fingerprint data D1 set in the latch circuit 41 of each of the matching units 40A to 40C; That is, in step SP112, the number of combinations that satisfy the relative positional relationship between the linear images cut out is detected.

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P118, where it is determined whether the number of detected combinations is equal to or greater than a predetermined value M. If a positive result is obtained here, the system control circuit 3 proceeds to step SP119, sets this combination of I, J, and K as a candidate specified by the variable Q, and then increments the variable Q in the subsequent step SP120. The process proceeds to step SP121.
On the other hand, if a positive result is obtained in step SP118, the system control circuit 3 proceeds to step SP11
8 directly proceeds to step SP121.

In this step SP121, the system control circuit 3 determines whether or not the coordinate value specified by the variable K is the last coordinate value detected by the collating unit 40C.
Here, if a negative result is obtained, the process proceeds to step SP122, where the variable K is incremented.
Return to 7. Accordingly, the system control circuit 3 sequentially switches the variable K with respect to the coordinate values (XIA, YIA) and (XJB, YJB), and these coordinate values (XIA, YIB).
A) Among the combinations of (XJB, YJB) and (XKC, YKC), candidates based on the variable Q are sequentially detected from the number of combinations that satisfy the relative positional relationship of the linear images.

When the variable K is sequentially switched to the last coordinate value in this way, the system control circuit 3 proceeds to step SP123 when a positive result is obtained in step SP121. Here, the system control circuit 3 determines whether or not the coordinate value specified by the variable J is the last coordinate value as in the case of the variable K. If a negative result is obtained here, the system control circuit 3 proceeds to step SP124.
Then, the variable J is incremented, the variable K is set to the initial value 1, and the process returns to step SP117.

Further, the variable J is incremented in this way, and this processing procedure is repeated, and the process proceeds to step SP123.
When a positive result is obtained in, the system control circuit 3
Moves to step SP125, determines whether the coordinate value designated by the variable I is the last coordinate value, and if a negative result is obtained here, moves to step SP126, increments the variable I and sets the variable J and K are initial value 1
After returning to step SP117, the process returns to step SP117.

In this way, the system control circuit 3 excludes, from the processing objects, combinations that do not satisfy the relative positional relationship from partial combinations from the coordinate values obtained by sequentially scanning the linear images. As a result, the fingerprint matching device 1 excludes combinations of coordinate values for which a high degree of similarity was detected by chance from candidates, and retains the remaining combinations as candidates specified by the variable Q.

The system control circuit 3 similarly detects a combination that satisfies the relative positional relationship between the candidate selected in this way and the coordinate values of the remaining collating units 40D to 40H, thereby reducing the collation rate. To detect. That is, the system control circuit 3 determines in step SP127 (FIG.
7) Set the variables Q, L, M, N, O, and P to the initial value 1. Here, the variables L, M, N, O, and P correspond to the matching units 40D to 40D.
40H is a variable for specifying the coordinate value stored in the coordinate group memory 49 for each.

Subsequently, the system control circuit 3 proceeds to step S
Moving to P128, where these variables Q, L, M, N,
Coordinate values (XQA, YQA) specified by O and P,
(XQB, YQB), (XQC, YQC), (XLD,
YLD), (XME, YME), (XNF, YNF),
Among the combinations of (XOG, YOG) and (XPH, YPH), the number of combinations that satisfy the relative positional relationship of the linear images is detected.

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P129, in which the number of combinations is recorded, and then proceeds to step SP130. Here, the system control circuit 3 determines whether or not the coordinate value designated by the variable P is the last coordinate value, and if a negative result is obtained here, proceeds to step SP131, increments the variable P, and proceeds to step SP127. Return.

On the other hand, if this processing procedure is repeated and a positive result is obtained in step SP130, the system control circuit 3 proceeds from step SP130 to step S130.
The process proceeds to P132, where it is determined whether the coordinate value designated by the variable O is the last coordinate value. If a negative result is obtained here, the process proceeds to step SP133, where the variable O is incremented and the variable P is initialized. The value is set to 1 and the process returns to step SP127.

This processing is repeated to return to step SP1
If a positive result is obtained in step 32, the system control circuit 3 proceeds from step SP132 to step SP134, where it determines whether or not the coordinate value specified by the variable N is the last coordinate value, and obtains a negative result here. Then, the process proceeds to step SP135, where the variable N is incremented, and the variables P and O are set to the initial value 1, and the process proceeds to step SP12.
Return to 7. If a positive result is obtained in step SP134, the system control circuit 3 proceeds to step SP13
Then, the process moves from step 4 to step SP136, where it is determined whether or not the coordinate value specified by the variable M is the last coordinate value. If a negative result is obtained here, the process proceeds to step SP137, where the variable M is incremented and P, O, and N are set to the initial value 1, and the process returns to step SP127.

Similarly, when a positive result is obtained in step SP136, the system control circuit 3
The process moves from step SP136 to step SP138, where it is determined whether or not the coordinate value designated by the variable L is the last coordinate value. If a negative result is obtained here, the process proceeds to step SP138.
137, the variable L is incremented, and the variables P, O, N, and M are set to the initial value 1, and the process proceeds to step SP1.
Return to 27. If a positive result is obtained in step SP138, the system control circuit 3 proceeds to step SP1.
38, the process proceeds to a step SP140 (FIG. 38), where it is determined whether or not the coordinate value designated by the variable Q is the last coordinate value. If a negative result is obtained here, a step SP14 is performed.
2, the variable Q is incremented, and the variables P, O, M, N, and L are set to the initial value 1, and the process proceeds to step SP
Return to 127.

As a result, the system control circuit 3 detects the number of combinations that satisfy the relative positional relationship for each combination of these coordinate values. At this time, the combinations that do not satisfy the relative positional relationship due to partial combinations are excluded. By doing so, this process is executed by narrowing down candidates in advance, and the time required for the process is shortened accordingly. That is, when a coordinate value is detected for each of the eight coordinate values, the combination of these coordinate values is the eighth power of the value a.

On the other hand, the number of combinations in the preliminary processing is the third power of a.
For example, when the candidates are narrowed down to one combination, the remaining combinations are a 5th power of a, so that the 5th power of a is much larger than the 3rd power of a, the combination required for the processing is approximately a can be reduced by the third power of a. That is, in this case, if a is set to the value 10, the time required for the processing can be reduced to almost 1/1000. This allows the fingerprint matching device 1 to execute the fingerprint registration process in a short time.

When the number of combinations satisfying the relative positional relationship is detected for each combination in this way, the system control circuit 3 detects the value of the maximum number of combinations in step SP142, and proceeds to step SP143. In, the value of the number of combinations is set as the collation rate, and the process returns to the main routine from step SP144. Thus, in this process, for each of the eight linear images, the coordinate values satisfying the relative positional relationship are stored in the coordinate group memory 49 with respect to the relative positional relationship when the eight linear images are cut out. When recorded, the system control circuit 3 detects the collation rate of 8/8 in a simplified manner.

Thus, the system control circuit 3 cuts one image into a line between fingerprint images taken a plurality of times from the same person and scans it on another image in a raster manner. The degree of similarity is comprehensively determined, and the matching rate is detected.

Subsequently, the system control circuit 3 proceeds to step S
The process moves to P145 (FIG. 35), where step SP111 is executed.
It is determined whether or not a matching rate has been detected for the rotated image generated in step SP1. If a negative result is obtained, the process proceeds to step SP1.
The process proceeds to 46, where the processing target is switched to the rotated image, and then returns to step SP112 to repeat the same processing.

As a result, the system control circuit 3
As shown in FIG. 40, a linear image is cut out from one fingerprint data stored in the rotation image memory 6B in the horizontal direction, and the fingerprint data of the cut out image is set in the matching units 40A to 40H, respectively. In addition, the corresponding fingerprint data D1 are simultaneously and parallelly checked in the raster scanning order by the collating units 40A to 40A.
40H, the previous linear image is raster-scanned on another image, and the degree of similarity is comprehensively determined in the overlapping portion to detect the matching rate.

In this manner, the system control circuit 3 cuts out a linear image substantially horizontally and vertically from the fingerprint image input from the fingerprint data input section 4, and separates the linear image into other horizontal and vertical directions. The degree of similarity to the image based on the fingerprint data is detected, and thus the degree of similarity in two dimensions is determined by combining the processing of one-dimensional image data that is relatively easy to process. .

When the matching rate is detected for the rotated image in this way, the system control circuit 3 proceeds to step SP1.
When a positive result is obtained in step 45, the process proceeds to step SP147, in which the matching rate having a value larger than the two matching rates is set as the correlation value of the n images for which the m images are to be compared. Return to the main routine.

As a result, the system control circuit 3 calculates correlation values for all combinations among a plurality of fingerprint images taken of the same person, selects fingerprint data having the highest correlation value, and performs fingerprint collation. And selectively register a convenient fingerprint image. Thereby, in the fingerprint collation device 1, the accuracy of fingerprint collation can be improved.

(1-4-5) Fingerprint Verification Processing In the state where the fingerprint data D1 has been registered in the fingerprint database 5 in this way, when the user operates the key input unit 2 to input a request for fingerprint verification, system control is performed. The circuit 3 executes the processing procedure shown in FIGS. 41 and 42 to perform fingerprint collation.

That is, the system control circuit 3 proceeds from step SP150 to step SP151, and
A message is displayed via (FIG. 1) to prompt the user to place the finger at the finger placement position. Next, the system control circuit 3
Moves to step SP152, and determines whether or not a finger has been placed at the finger placement position. If a negative result is obtained here, step SP152 is repeated.

In step SP152, the system control circuit 3 inputs the output data of the serial / parallel conversion circuit 21 via the data bus BUS (FIG. 2), and sets the logical level of the output data for a predetermined segment. By counting, it is detected whether or not a finger is placed on the basis of the output data. This counting process is executed by counting the number of bits of the logical L level for a predetermined plurality of segments corresponding to substantially the center portion of the finger placement position, and determining whether the count value has exceeded a predetermined value. Thus, the system control circuit 3 starts the fingerprint collation processing by using the imaging result as a trigger, thereby simplifying the operation of the fingerprint collation device 1 and improving the usability.

If an affirmative result is obtained in step SP152, the system control circuit 3 proceeds to step SP152.
The process proceeds to 153, where it is determined whether or not a biological reaction is detected by the pulse wave detector 22. Here, if a negative result is obtained, the system control circuit 3 proceeds to step SP154 and ends this processing procedure, whereas if a positive result is obtained in step SP153, the system control circuit 3 proceeds to step SP155.

In this step SP 155, the system control circuit 3 inputs the fingerprint data D 2 to be inspected outputted from the fingerprint data input unit 4 to the inspection fingerprint memory 6. Thus, the system control circuit 3 performs fingerprint collation only when a biological reaction of a finger is detected, thereby improving security.

Subsequently, the system control circuit 3 proceeds to step S
The process moves to P156, and a comparison determination process is executed. Here, in the present embodiment, the system control circuit 3 uses the user ID input via the key input unit 2 as a reference, and the fingerprint data D1 registered for the corresponding user ID.
And the fingerprint data D2 input from the fingerprint data input unit 4.
Is determined as to whether the fingerprints match. Further, at this time, when a plurality of fingerprint data D1 are registered for the same user ID in the fingerprint database 5, it is determined whether or not the fingerprint data D1 has the highest priority.

Subsequently, the system control circuit 3 proceeds to step S
The process shifts to P157, where it is determined whether a match determination result is obtained. If a positive result is obtained, step SP158 is performed.
And outputs the result of the determination of the match, and then proceeds to step SP1.
Move to 59. Here, the system control circuit 3 records the collation rate in the fingerprint database 5.

Subsequently, the system control circuit 3 proceeds to step S
The process proceeds to P160 (FIG. 42), and the change in the collation rate recorded in the fingerprint database 5 is confirmed. Here, when the collation rate is gradually reduced to a certain value or less, it is determined that the change in the collation rate has exceeded an allowable range, and the system control circuit 3
Moves to step SP161, prompts the user to update and register the fingerprint data, and then moves to step SP162 to end this processing procedure.

That is, in this embodiment, the degree of similarity between the fingerprint data D2 to be inspected and the fingerprint data D1 registered in the fingerprint database 5 is measured, and the degree of similarity is represented by the collation rate. In addition, the fingerprint data D2 is determined based on whether or not the collation rate exceeds a predetermined reference value.
It is determined whether the fingerprint according to the fingerprint data coincides with the fingerprint according to the fingerprint data D1.

However, in a child or the like, the finger may become larger as it grows up. In this case, the change in the size of the finger causes the collation rate to gradually decrease as shown in FIG. Become. As a result, in this embodiment, when the matching rate gradually decreases and becomes equal to or less than the fixed value TH,
The user is set to register the fingerprint data D2 again, whereby the collation rate is restored and the fingerprint collation can be performed reliably.

If such a change in the collation rate is not observed, the system control circuit 3 executes step SP
When a positive result is obtained in 160, the process directly proceeds from step SP160 to step SP162, and this processing procedure ends.

On the other hand, if a negative result is obtained in step SP157 (FIG. 41), the system control circuit 3 moves to step SP163 and determines whether or not the frame has already been moved. Here, in this embodiment, FIG.
As shown in FIG. 4, the fingerprint image based on the fingerprint data D1 to be inspected is cut out linearly within a predetermined frame to detect the collation rate. As a result, in the comparison determination process in step SP156, if no match result is obtained, the system control circuit 3 proceeds to step S156.
The process moves to step SP164 from P163, and moves this frame. Further, from step SP164 to step SP15
Returning to 6, the comparison determination processing is executed based on the moved frame.

Thus, when the user registered in the fingerprint database 5 performs fingerprint collation, the system control circuit 3 executes a fingerprint collation process so as to reliably obtain a match determination result. It is designed to improve the accuracy of matching.

If no match is obtained even when the frame is moved in this way, the system control circuit 3 obtains a negative result in step SP157 and then obtains a positive result in step SP163. Moves to step SP165 (FIG. 42). Here, the system control circuit 3 sends another fingerprint data D1 to the same person.
It is determined whether or not is registered.

As described above with reference to FIG. 15, when the fingerprint data D1 is registered in the fingerprint database 5 for the index finger, the middle finger, and the little finger together with the priority data, the system control circuit 3 proceeds from step SP165 to step SP166. Then, after switching the fingerprint data D1 of the matching partner in accordance with the priority order, step SP1
Return to 55. As a result, the system control circuit 3 first compares the fingerprint data D2 input via the fingerprint data input unit 4 with the fingerprint data D11 of the index finger registered in the fingerprint database 5, and then performs the middle finger operation. The fingerprint collation processing is sequentially performed between the fingerprint data D12 of the small finger and the fingerprint data D13 of the little finger. For example, when a user who cannot hover at the finger placement position due to injury of the index finger places the middle finger However, it is possible to reliably determine that the person is the person. Further, at this time, by sequentially switching the fingerprint data D1 in accordance with the priority order, the time required for collation is shortened accordingly.

In the case where it is difficult to obtain a match determination result even when the fingerprint data D1 is switched, the system control circuit 3 proceeds to step SP167 because a negative result is obtained in step SP165. Here, the system control circuit 3 determines whether or not the collation rate is equal to or less than a certain value. If an affirmative result is obtained, the process proceeds to step SP168, outputs a mismatch determination result, and proceeds to step SP162. This allows the system control circuit 3 to output a mismatch determination result when it can be reliably determined that the user is different.

As described above with reference to FIG. 43, in the case of a child or the like, the collation rate may decrease as the child grows up. Depending on the processing of SP160, it is difficult to recover the collation rate, and eventually it is difficult to obtain a match determination result. Further, even when the finger is injured and the fingerprint is damaged, it is difficult to obtain a matching determination result.

In this case, in the collation rate, an extreme decrease in the collation rate as in the case of collation with another person's fingerprint is not observed, and a negative result is obtained in step SP167. Further, in such a case, there is a characteristic that a substantially same collation rate can be obtained by repeatedly collating with the fingerprint data D1.

In this embodiment, if a negative result is obtained in step SP167, the system control circuit 3 proceeds to step SP169 to determine whether or not a series of fingerprint collation is repeated for the same user ID. When a negative result is obtained, step SP151 is performed.
Return to Thus, when it is difficult to reliably determine the match or the mismatch, the system control circuit 3 fetches the fingerprint data D2 again and executes the fingerprint collation processing. This allows the user to place the middle finger instead of the index finger, for example, and perform fingerprint collation again as needed. )

In this fingerprint collation process again, if a negative result is obtained again in step SP167,
When a positive result is obtained in step SP169, the system control circuit 3 proceeds to step SP170, where the criterion for the match determination is reduced, and then returns to step SP155. That is, when it is difficult to judge complete match or mismatch by repeating such processing, and when the collation rate is kept within a certain range, it is possible to judge that they match and obtain a correct judgment result. The circuit 3 is configured to be able to reliably output a match determination result for a registered user.

FIG. 45 is a flowchart showing the fingerprint data input processing. The system control circuit 3 determines in step S
The process moves from step P171 to step SP172, where an effective area detection process is executed. Here, this effective area detection processing is executed according to the same processing procedure as the effective area detection processing described above with reference to FIG. 25, whereby the system control circuit 3 activates an area available for fingerprint collation in units of segments. An area is set, and thereafter, various processing is executed based on the effective area, so that the various processing can be simplified.

Subsequently, the system control circuit 3 proceeds to step S
The process moves to P173, where a finger position detection process is executed. Here, this finger position detection processing is executed by the same processing procedure as the finger position detection processing described above with reference to FIG. 27, and the system control circuit 3 performs fingerprint collation only when the user has correctly placed the finger, It has been made to improve the accuracy of the.

Subsequently, the system control circuit 3 proceeds to step S
The process moves to P174, where the zoom process is executed. FIG.
As shown in (2), in this zoom processing, the system control circuit 3 determines from step SP175 to step SP176
Then, the scaling factor of the corresponding fingerprint data D1 is loaded from the fingerprint database 5, and in the following step SP177, the thinning rate of the thinning circuit 20 is set according to the loaded scaling factor, and the process shifts to step SP178 to return to the main routine. . Thus, the system control circuit 3 takes in the fingerprint data D2 to be inspected based on the magnification of the fingerprint data D1 registered in the fingerprint database 5, and inputs the fingerprint data D2 under the same conditions as those at the time of registration. The accuracy of fingerprint collation is improved accordingly.

Subsequently, the system control circuit 3 determines in step S
The process shifts to P179 (FIG. 45), where the threshold correction process is executed. Then, the process shifts to step SP180 to execute the blackout process, and then executes the angle detection process in step SP181. Further, the system control circuit 3 executes the following step S
In P182, the fingerprint data D is stored in the erect image memory 6A.
After inputting 2, the process moves to step SP183. Here, this threshold value correction processing, blackout processing, and angle detection processing
29, 30, and 31 are executed by the same processing procedure as the corresponding processing described above with reference to FIGS. Accordingly, the system control circuit 3 performs the comparison circuit 14 so that the fingerprint data D2 to be inspected can be surely subjected to fingerprint collation.
The reference voltage is corrected, the image quality is effectively prevented from being degraded due to dirt on the finger, and the fingerprint data D2 is taken in with the inclination of the finger within a certain range.

At step SP183, the system control circuit 3 determines whether or not the comparison circuit 14 of the fingerprint data input unit 4 has set the reference voltage REF2. If a negative result is obtained here, the process proceeds to step SP184. After switching the reference voltage, the process returns to step SP182. As a result, the system control circuit 3 determines in step SP1
While the processing procedure of 82-SP183-SP184-SP182 is repeated and the reference voltage is corrected by the threshold correction processing, the reference voltage that is the setting reference of the threshold correction processing is sequentially updated, and a total of three reference voltages are set. After storing the fingerprint image in the inspection fingerprint memory 6, the process moves to step SP185, and this processing procedure ends.

FIGS. 47 and 48 are flowcharts showing the procedure of the comparison / determination process. System control circuit 3
Moves from step SP190 to step SP191 in this processing procedure, where the fingerprint data D2 stored in the erect image memory 6A is transferred to the image rotation circuit 31 in segment units, and is output from the image rotation circuit 31. The fingerprint data D2 is stored in the rotation image memory 6B.
Further, the corresponding fingerprint data D1 of the fingerprint database 5 is transferred to the image rotation circuit 31 in segment units, and the fingerprint data D1 output from the image rotation circuit 31 is stored in the database rotation image memory 5B. Thus, in this step SP191, the system control circuit 3 rotates the image to be processed by 90 degrees (see FIGS. 10 and 1).
1).

Subsequently, the system control circuit 3 proceeds to step S
Moving to P192, as shown in FIG. 49, the frame is set by tilting the frame by the tilt Δθ detected in the above-described angle detection process (FIG. 45, step SP181).
Correct the tilt of the finger and set the frame.

That is, as shown in FIG. 50, the system control circuit 3 stores the data of 8 pixels in the test fingerprint memory 6 as the 1-byte fingerprint data D2. In units, control the address of the memory control circuit 30;
Approximately, a linear image (64 pixels) inclined from the horizontal direction by an angle Δθ is sequentially cut out from the upper direction of the frame.

Further, the system control circuit 3 outputs the fingerprint data D2A for forming the linear image thus cut out.
To D2H are output to the matching units 40A to 40B, respectively.
The data is stored in the latch circuit 41 of each of the matching units 40A to 40B (FIG. 17). Accordingly, the system control circuit 3 sets the linear image to be inclined with respect to the image based on the fingerprint data D1, and corrects the inclination of the fingerprint data D2 image with respect to the fingerprint data D1 image. Thereby, in the fingerprint matching device 1,
Even if the user tilts his / her finger within a certain range,
The fingerprint collation can be performed reliably.

When the frame is set as described above and a linear image is cut out, the system control circuit 3 determines whether or not the black-out segment detected by the black-out processing (FIG. 45, step SP180) A predetermined judgment procedure is executed so as not to cut out the image of the fingerprint data D2.
A to D2H are set, thereby effectively preventing the accuracy of fingerprint collation from lowering. In addition, a frame is set on the basis of the effective area detected in advance, thereby also omitting useless matching rate detection processing and improving the accuracy of fingerprint matching.

Further, in advance, the number of times the logical level is switched is counted for the fingerprint data D2A to D2H cut out linearly, and the portion where the count value is equal to or less than the predetermined value is excluded from the target. As a result, the linear cut-out portion crosses the fingerprint for a predetermined number or more, so that the cut-out region sufficiently contains significant information for fingerprint matching, and this also improves the accuracy of fingerprint matching. Has been made to be able to improve. Incidentally, the system control circuit 3 determines the position of the frame as in the case of fingerprint registration.

Subsequently, the system control circuit 3 proceeds to step S
In P194, the fingerprint data D1 is sequentially transferred from the fingerprint database 5 to the matching units 40A to 40H so that the fingerprint data D1 is successively arranged in the order of raster scanning. As a result, the system control circuit 3 makes each of the matching units 4 as shown in FIG.
In 0A to 40H, the linear image is scanned in the raster scanning order on the image based on the fingerprint data D1, and the number of matching bits between the linear image and the image based on the fingerprint data D1 overlapping each other is checked. It is detected by the counter 45 of the units 40A to 40H (FIG. 16). Further, the system control circuit 3 determines the coordinate value at which the count value rises above the threshold value set in the register 46, that is, the position at which the two overlapping images are determined to be extremely similar.
The coordinate data of the fingerprint data D1 is sequentially taken into the coordinate group memory 49 and held.

When the fingerprint data D1 is output from the fingerprint database 5, the system control circuit 3
The operation of the comparison circuit 44 is switched so that the comparison result is output at a certain logical level for the corresponding bit of the fingerprint data D1 determined to be an invalid area or a blackout condition. Is changed by that amount, and the determination criterion in the comparison circuit 47 is changed. As a result, the system control circuit 3 masks such invalid areas and black-out areas, and processes these areas.

The system control circuit 3 supplies fingerprint data from the fingerprint database 5 so that the linear image scans the effective area, thereby omitting unnecessary processing for fingerprint collation and simplifying the entire processing. Become

Subsequently, the system control circuit 3 determines in step S
The process moves to P195, where a matching ratio detection process is executed. Here, this collation rate detection processing is performed in accordance with FIGS.
Is performed by the same process as the above-described matching ratio detection process.

Thus, the system control circuit 3 converts the linear image cut out from the fingerprint data D2 into the fingerprint data D1.
, The degree of similarity of the overlapping portion is comprehensively determined in a state where the image is raster-scanned on the image according to the above, and the matching rate is detected. At this time, the system control circuit 3 forms a partial combination from the coordinate values obtained by sequentially scanning the linear image, and selects in advance a combination of coordinate values where a high degree of similarity is detected by chance. Then, the number of combinations that satisfy the relative positional relationship with the remaining coordinate values is detected. As a result, the system control circuit 3 shortens the time required for the processing, and performs the fingerprint collation in a short time.

Subsequently, the system control circuit 3 determines in step S
The process moves to P196, where it is determined whether or not the angle θ is the last.
Here, the system control circuit 3 determines the tilt correction angle Δ
As shown in FIG. 53, the inclination of the image cut out linearly is changed around θ, and the collation rate is detected a plurality of times. Determine the fingerprint match. In this case, the system control circuit 3
Moves to step SP197 because a negative result is obtained in step SP196, and after updating the angle θ, returns to step SP193.

Thus, the system control circuit 3 sequentially switches the angle, cuts out a linear image from the image based on the fingerprint data D2, and sequentially detects the collation rate from the cut out image. Thus, the system control circuit 3 can surely perform fingerprint collation even in a case where the inclination is difficult to be completely corrected by the angle correction and a part of the fingerprint is deformed. In this embodiment, the collation rate is detected at each of five angles around the angle Δθ.

When the collation rate is detected for each angle as described above, a positive result is obtained in step SP196, and the system control circuit 3 proceeds to step S196.
Move to P198. Here, the system control circuit 3 determines whether or not the frame switching has been completed. That is, FIG.
As shown in FIG. 5, in this embodiment, three frames are set so as to be continuous, and the coincidence of the fingerprint is determined by each frame, thereby improving the discriminating power.

In this case, the system control circuit 3 obtains a negative result in step SP198,
The process moves to step SP199, where the frame is switched from frame 1 to frame 2 and then to step SP193. As a result, the system control circuit 3 similarly detects the collation rate for the frame 2 and then detects the collation rate for the frame 3 in the same manner.

When the collation rate is detected for frame 3 in this manner, a positive result is obtained in step SP198, and the system control circuit 3 proceeds to step SP2.
Move to 00 (FIG. 48). Here, the system control circuit 3
It is determined whether or not the same processing has been performed on the rotated image stored in the rotated image memory 6B and the rotated image memory for database 5B. In this case, a negative result is obtained, and the process proceeds to step SP201. Here, the system control circuit 3 operates the upright image memory 6A to be processed up to that point.
And the fingerprint data D1 and D2 stored in the rotation image memory 6B and the database rotation image memory 5B instead of the fingerprint data D1 and D2 held in the fingerprint database 5.
After switching the processing target to 2, the process returns to step SP192.

In this manner, the system control circuit 3 similarly detects the collation rate by sequentially switching the angle in the state where the angle is corrected and sequentially switching the frame also for the image rotated by 90 degrees. As a result, the fingerprint collation device 1 can perform fingerprint collation more reliably. Thus, the system control circuit 3 cuts out a linear image in a substantially horizontal direction and a vertical direction from the fingerprint image input from the fingerprint data input unit 4, and stores the horizontal and vertical directions in the fingerprint database. The degree of similarity to the fingerprint image is detected, and in this case, two-dimensional similarity is determined by combining one-dimensional image data processing that is relatively easy to perform.

When the collation rate is detected for the rotated image in this manner, a positive result is obtained in step SP200, and the system control circuit 3 proceeds to step S200.
Move to P202. Here, the system control circuit 3 determines whether or not the detection processing of the collation rate has been completed for all the images taken into the test fingerprint memory 6 by switching the reference voltage. If a negative result is obtained here, step SP203
And after switching the image to be processed, step SP
Return to 103.

Thus, the system control circuit 3 switches the frame and the inclination of each of the three fingerprint images captured by switching the reference voltage, and detects the collation rate of the erect image and the rotated image.

When the collation rate is detected in this manner, the system control circuit 3, when a positive result is obtained in step SP202, proceeds to step SP204, where the collation rate equal to or more than a predetermined value is set in each frame of the erect image. Is determined. If a positive result is obtained here, the system control circuit 3 proceeds to step SP205, and similarly determines whether or not a collation rate equal to or greater than a predetermined value is obtained in each frame of the rotated image.

If a positive result is obtained here, the system control circuit 3 proceeds to step SP206, and loads, from the coordinate group memory 49, the coordinate values at which the matching ratio equal to or greater than the predetermined value is detected in the erect image and the rotated image. I do. Further, it is determined whether or not the coordinate value is below a predetermined allowable range. That is, as shown in FIGS. 55 and 56, the fingerprint data D1 and D2
When the fingerprints match, in the erect image and the rotated image, for example, the linear image D2A is scanned, and the coordinate values X1, Y1 and X2, Y2 at which such a matching ratio equal to or more than a predetermined value is detected are: The erect image and the rotated image have the initial relative positional relationship when the frame is set. Incidentally, when the frame 1 is set at the same position in the erect image and the rotated image with respect to the image based on the fingerprint data D2, the X coordinate value and the Y coordinate value have a relationship of X1 = Y2 and Y1 = X2. .

When this relationship is disturbed, when a line image is accidentally scanned in the horizontal or vertical direction,
It can be determined that another person has a similar fingerprint pattern locally. Thus, in this embodiment, the system control circuit 3 further improves the discriminating power based on the coordinate values based on the erect image and the rotated image.

If a positive result is obtained in step SP206, the system control circuit 3 proceeds to step SP20.
Move to 7. Here, in the erect image, the system control circuit 3 determines whether or not the coordinate values detected in the adjacent frames are within a certain allowable range. It is determined whether or not the relative positional relationship is satisfied.

That is, as shown in FIG. 57, when the fingerprints of the fingerprint data D1 and D2 coincide with each other, the linear image D2A is scanned for each frame as described above, and the coordinates of the fingerprint data D1 are used to obtain a constant collation rate. Values X1, Y1 and X
2, Y2 is the initial relationship when each frame is set. If this relationship is disturbed, and if a line image is accidentally scanned in the horizontal direction, it can be determined that the person has a similar fingerprint pattern locally. Thus, in this embodiment, the system control circuit 3 further improves the discriminating power based on the coordinate values between the adjacent frames.

When a positive result is obtained between adjacent frames for the erect image in this manner, the system control circuit 3
Performs the same determination process on the adjacent frame also for the rotated image, thereby further improving the discriminating power.

In the processing of each of the steps SP204 to SP207, the system control circuit 3 determines whether or not a positive result is obtained for each fingerprint image whose angle and reference voltage have been switched. If a positive result is obtained in the image, the process proceeds to the next step. Thereby, the system control circuit 3 determines whether or not the fingerprint image whose angle and reference voltage have been switched matches the image based on the fingerprint data D1 by so-called OR determination. As a result, the system control circuit 3 can effectively avoid a decrease in the discriminating power between the person and the other person, and can surely output the matching result when the person himself / herself.

When a positive result is obtained in step SP207 in this way, the system control circuit 3 proceeds from step SP207 to step SP208. Only in this case, after determining that the fingerprints match, the system control circuit 3 proceeds to step SP208.
From 209, the process returns to the main routine.

As a result, the system control circuit 3 determines that the fingerprints match only when the match can be reliably determined.
This improves the accuracy of fingerprint collation. If necessary, the processing described above with reference to FIG. 42 is repeated to output a match determination for a user who should be determined to be a match, and to output a mismatch determination for a user who should be determined to be a mismatch. Has been made.

The updating process of the criterion described above with reference to FIG.
The judgment criteria of P207 will be updated. Similarly, the recording of the collation rate in step SP159 in FIG. 42 and the determination of a change in the collation rate in step SP160 are performed by recording the collation rates detected by a plurality of frames and inclinations in this manner.
Also, the recorded collation rate is determined. Further, the determination of the collation rate in step SP166 repeats the determination in steps SP204 to SP207 described above.

(2) Operation of Embodiment In the above configuration, when the power supply is turned on (FIGS. 2, 3 to 6, and 20), the fingerprint matching device 1 The reference data of the light amount correction memory 16 set at the time of shipment from the factory is set in the reference voltage generation circuit 15, whereby the reference voltage of the comparison circuit 14 is set so as to correct the light amount unevenness of the optical system in units of segments. . As a result, the deterioration of the fingerprint collation accuracy due to the uneven light amount of the optical system can be effectively avoided, and the fingerprint collation can be performed with high accuracy by the optical system having a simple configuration.

In this state, the illumination light emitted from the light source 12 is reflected by the bottom of the isosceles triangular prism 11 and
An image is picked up by the camera 13 and the image pickup result is
4 is binarized. The binarized signal S1 output from the comparison circuit 14 is latched by the latch circuit 18 and converted into 1-bit image data, and then thinned out by the thinning circuit 20.
1 converts the data into 8-bit image data in units of 8 pixels.

The image data output from the serial / parallel conversion circuit 21 is input to the system control circuit 3 via the data bus BUS. Here, the bit whose logical level is falling is set in units of segments. The number is counted (FIG. 21), and the abnormality of the optical system is monitored. As a result, dirt on the bottom surface of the isosceles triangular prism 11 at the finger placement position, deterioration of the light source 12, and the like are detected, and maintenance processing is executed as necessary, thereby effectively avoiding deterioration of fingerprint collation accuracy.

When the user operates the key input unit 2 (FIG. 1) in this state, the fingerprint matching device 1 executes the corresponding fingerprint registration process and fingerprint matching process (FIG. 20).

In the fingerprint registration processing, the fingerprint collation apparatus 1 prompts the user to place the index finger through the display unit 8 (FIG. 22), and the change of the image data output from the serial / parallel conversion circuit 21 is When a predetermined segment is detected, a fingerprint registration process is started, whereby the fingerprint can be registered by a simple operation.

Subsequently, a pressure sensor 2 arranged beside the finger
According to 3 (FIG. 2), the biological reaction of the finger is detected. If the biological reaction is not detected, the fingerprint registration processing is stopped.
As a result, security is improved in terms of fingerprint registration.

On the other hand, when a biological reaction is detected, the conditions for inputting fingerprint data are adjusted, and a plurality of fingerprint images are fetched. Among them, the image most suitable for fingerprint collation is registered in the fingerprint database 5.

At the time of this registration, when the registration of the index finger is completed, the fingerprint collation device 1 subsequently proceeds to the fingerprint data D1 of the middle finger.
, And then the fingerprint data of the little finger is registered, and the index finger, the middle finger, and the little finger are registered in order of priority (FIG. 15). Thus, in the fingerprint collation device 1, in the fingerprint collation process, any one of the index finger, the middle finger, and the little finger can perform fingerprint collation, and the usability can be improved accordingly. At this time, fingerprint collation processing is executed in accordance with the priority order, so that the time required for collation can be reduced.

In the actual processing of the image pickup result (FIG. 23), the fingerprint collation apparatus 1 counts the output data of the serial / parallel conversion circuit 21 by the system control circuit 3 so that the region where the background is imaged becomes a segment. Removal is performed in units, and thereby an effective area in which a fingerprint is actually imaged is detected (FIG. 25). As a result, a series of subsequent processing is executed based on the effective area, and the time required for the processing is shortened accordingly.

Subsequently, it is determined whether or not the placement position of the finger is correct based on the number of segments in the effective area (FIG. 41), and deterioration of the matching accuracy due to improper placement of the finger is effectively avoided. In the subsequent zoom processing, the thinning circuit 20 (FIG. 2) is performed based on the number of segments in the effective area.
In this case, the thinning-out rate is changed, and thus, even if the child's finger or the like is picked up, an image is taken at a magnification suitable for fingerprint matching, and the accuracy of fingerprint matching is improved.

In the subsequent threshold value correction process (FIG. 29), the number of bits of logic H is detected for each segment of the effective area, and a finger is placed on the bottom surface of the isosceles triangular prism 11 for the corresponding segment. Is detected, and the count value is 30 to 7 in each segment.
The threshold value correction memory 2 is set so as to fall within the range of 0 [%].
4 (FIG. 2) is corrected. The contents of the threshold value correction memory 24 are set so as to correct the reference data of the light amount correction memory 16, thereby deteriorating the fingerprint collation accuracy, which changes with the pressing force, in units of segments, and the isosceles triangular prism 11. Deterioration of fingerprint collation accuracy due to dirt or the like is effectively avoided.

Subsequently, in the blackout condition processing (FIG. 3
0) By detecting the number of bits of logic L for each segment of the effective area, a blackout segment in which it is difficult to secure a ratio of a light portion to a dark portion of a predetermined value or more is detected, Are excluded from matching. It is also noted that the finger is abnormally wet. As a result, the imaging result is processed under conditions suitable for fingerprint matching, and the accuracy of fingerprint matching is improved.

In the following angle detection processing (FIG. 31,
32 and 33), the inclination of the finger is detected based on the segment of the effective area, and when the inclination is abnormal, the user is alerted. As a result, the deterioration of the fingerprint collation accuracy due to the tilted placement of the finger can be effectively avoided.

When the input conditions of the fingerprint data D1 are adjusted in this way (FIG. 24), the fingerprint collation device 1 outputs the fingerprint data D1 output from the serial / parallel conversion circuit 21.
After 1 has been loaded into the erect image memory 6A of the N-piece inspection fingerprint memory 6 (FIG. 9), the reference voltage REF of the comparison circuit 14 is subsequently offset to the positive side by a predetermined voltage (FIG. 8).
N pieces of fingerprint data D1 are similarly taken into the erect image memory 6A, and conversely, the reference voltage REF is offset to the negative side by a predetermined voltage, and N pieces of fingerprint data D1 are
Is taken into the erect image memory 6A.

By switching the reference voltage REF by offsetting, in the fingerprint collation apparatus 1, the fingerprint image is made almost equal in width. As a result, even with a simple process of binarizing the imaging result and processing, the change of the binarized signal S1 due to the change in the pressing force, the distortion of the fingerprint, the contamination of the optical system for imaging the finger, etc. is absorbed, and the fingerprint collation is performed. A decrease in accuracy is effectively avoided.

In the fingerprint collation apparatus 1, the fingerprint image most suitable for fingerprint collation is selected from the 3N fingerprint images, and the selected fingerprint image is registered in the fingerprint database 5. Thereby, in the fingerprint matching device 1, the accuracy of fingerprint matching is improved.

The most suitable fingerprint image for the fingerprint collation is selected (FIG. 34). Assuming that one of the 3N images is fingerprint data registered in the fingerprint database 5, the fingerprint image is compared with other images. The matching process is repeated to detect the matching rate. Further, the same processing is repeated by sequentially switching the images assumed to be registered, and the fingerprint image having the largest correlation value, which is the collation rate, is selected and executed. This allows the fingerprint collation device to use the criterion corresponding to the actual fingerprint collation,
The fingerprint data D1 to be registered in the fingerprint database 5 is selected, and the fingerprint can be surely collated accordingly.

Specifically, in the fingerprint collation apparatus 1, the fingerprint data (FIG.
5, corresponding to n images) and fingerprint data for detecting the matching rate (m image corresponds to FIG. 35)
Respectively from the erect image memory 6A.
1, where the arrangement within the segment is changed, and fingerprint data is generated by rotating the segment by 90 degrees (FIG. 12). Further, the generated fingerprint data is stored in the rotation image memory 6B and the rotation image memory for database 5B, whereby an erect image and a rotated image obtained by rotating the erect image by 90 degrees are generated.

Further, in the erect image, eight linear images composed of image data of 64 pixels in the horizontal direction are sequentially cut out from the m image in the vertical direction, and each linear image data is collated. This is set in the latch circuits 41 of the units 40A to 40B (FIG. 16). Also, so that the fingerprint data of the n images is continuous in the raster scanning order,
The two pieces of image data are supplied to the collating units 40A to 40B in a simultaneous and parallel manner.

Further, the number of matching bits is counted by the counter 4
It is counted by 5 and it is judged by the comparing circuit 47 whether or not it exceeds a certain reference value. Further, when the value exceeds the predetermined value, the fingerprint data of the n images is compared with the collation units 40A to 40B.
The coordinate values of the corresponding n images are stored in the coordinate group memory 4 with reference to the address data AD of the memory control circuit 30 output to the memory 4.
9 recorded.

The eight linear images are raster-scanned on each of the n images (FIG. 18), and the degree of similarity is sequentially detected between the overlapping images, and the coordinate values of positions similar to each other by a certain value or more are detected. Is stored in the coordinate group memory 49. At this time, in the fingerprint collation device 1, the eight linear images are scanned by the eight collation units 40A to 40H simultaneously and in parallel, so that an image suitable for fingerprint collation can be selected in that short time. It has been made like that.

Further, when a linear image is cut out and set in the collating units 40A to 40H in this manner, a predetermined judgment procedure is executed so as not to cut out the linear image for a black-out segment. As a result, the fingerprint data D1 is set, whereby a decrease in fingerprint collation accuracy is effectively avoided. Also, for the effective area detected in advance, a frame for cutting out this linear image is set, so that a necessary image can be cut out in a short time, and further unnecessary processing can be omitted. I have. Further, for the fingerprint data cut out in a linear manner, the switching of the logic level is counted, and the portion where the count value is equal to or less than the predetermined value is excluded from the target, so that the portion where the significant information is sufficiently included in the fingerprint collation is included. It is cut out, and the accuracy of fingerprint collation can be improved also by this.

On the other hand, in the fingerprint data supplied in the raster scanning order, the operations of the comparison circuits 44 and 47 are switched so as to mask the blackened segments and the blank region segments. The reduction of the matching rate due to the region and the segment is effectively avoided.

When the coordinate data of the erect image is stored in the coordinate group memory 49, the same processing is subsequently performed using the fingerprint data stored in the rotated image memory 6B and the rotated image memory for database 5B ( FIG.
0), coordinate data is also recorded for the rotated image (FIG. 19).

The coordinate data recorded in this manner is:
The number of combinations satisfying the relative positional relationship of the linear images is detected for each combination of coordinate values by the matching units 40A to 40H by the matching ratio detection processing by the system control circuit 3 (FIGS. 36 and 37). The collation rate is detected based on the detected number.

At this time, the system control circuit 3 detects the number of combinations satisfying the relative positional relationship of the corresponding linear images for each combination of coordinate values obtained by the collation units 40A, 40B, 40C. Then, coordinate values where a high degree of similarity is detected by chance on the basis of the number of combinations are excluded from the inspection target, and the matching ratio is detected by the remaining combinations. As a result, the number of processing targets can be reduced from a partial combination in advance, the matching rate can be detected in a correspondingly short time, and an image suitable for registration can be selected in a short time.

Assuming that the n images have been registered in the fingerprint database in this way, when the matching ratio is detected for the m images, the matching ratio is similarly detected by switching the m images, and the matching ratio is detected for the 3N images. When the processing is completed, the n images are switched, the same processing is repeated, and the matching rate is detected for each combination.

The collation rate detected in this way is processed as a correlation value at the time of fingerprint registration, and the fingerprint data obtained assuming the largest correlation value and registered in the fingerprint database is the fingerprint data. It is registered in the database 5, thereby improving the accuracy of fingerprint collation.

On the other hand, in the fingerprint collation processing, the fingerprint collation apparatus 1 prompts the user to place his / her finger through the display unit 8 (FIG. 41), and changes in the image data output from the serial / parallel conversion circuit 21. As a result, the fingerprint collation process is started, whereby the fingerprint collation can be performed by a simple operation.

Subsequently, a pressure sensor 2 arranged beside the finger
According to 3 (FIG. 2), the biological reaction of the finger is detected. If the biological reaction is not detected, the fingerprint collation processing is stopped.
As a result, security is improved in terms of fingerprint collation.

On the other hand, when a biological reaction is detected, the fingerprint data input conditions are set, and then the fingerprint data is fetched. The corresponding user is compared with the fingerprint data registered in the fingerprint database 5. Then, the captured fingerprint data is subjected to fingerprint collation.

That is, also under the input condition of the fingerprint data at the time of the fingerprint collation (FIG. 45), an effective area is detected from the output data of the serial / parallel conversion circuit 21 (FIG. 25). As a result, a series of subsequent processes are executed based on the effective area, and the time required for collation is shortened accordingly. Also, it is determined whether or not the placement position of the finger is correct based on the number of segments in the effective area (FIG. 41), and deterioration of the matching accuracy due to improper placement of the finger is effectively avoided.

In the following zoom processing (FIG. 46), the thinning circuit 20 in the fingerprint data input unit 4 corresponds to the magnification registered in the fingerprint database 5.
Is set so that fingerprint data can be input at the same magnification as that at the time of fingerprint registration, thereby improving the accuracy of fingerprint collation.

In the threshold value correction processing (FIG. 2
9) The contents of the threshold value correction memory 24 (FIG. 2) are corrected for each segment of the effective area, thereby deteriorating the fingerprint collation accuracy that changes with the pressing force in units of segments, isosceles triangular prism Deterioration of fingerprint collation accuracy due to dirt or the like is effectively avoided. Further, in the blackout condition processing (FIG. 30), a case is detected in which a blackout segment that is inappropriate for fingerprint matching is detected, this segment is excluded from the matching target, and the finger is abnormally wet.
As a result, the imaging result is processed under conditions suitable for fingerprint matching, and the accuracy of fingerprint matching is improved.

In the following angle detection processing (FIG. 3
1, FIG. 32, FIG. 33), the inclination of the finger is detected with reference to the effective area, and when the inclination is abnormal, the user is warned. (FIG. 47, step SP192) is executed, thereby effectively avoiding deterioration of the fingerprint matching accuracy due to the tilted and placed finger.

When the input conditions for the fingerprint data D1 are adjusted in this way, in the fingerprint collation apparatus 1, the fingerprint data D2 output from the serial / parallel conversion circuit 21 is stored in the erect image memory of the inspection fingerprint memory 6 (FIG. 9). 6A, the reference voltage REF of the comparison circuit 14 is offset to the positive side by a predetermined voltage (FIG. 8), and the fingerprint data D2 is similarly captured in the erect image memory 6A, and vice versa. The reference voltage REF is offset to the negative side by a predetermined voltage, and the fingerprint data D2 is taken into the erect image memory 6A.

By switching the reference voltage REF by offsetting, in the fingerprint collation apparatus 1, the fingerprint image to be inspected is also made substantially equal in width, thereby changing the pressing force and the fingerprint collation accuracy due to the distortion of the fingerprint. Is effectively avoided.

The fingerprint data D2 thus captured into the inspection fingerprint memory 6 (FIG. 47) is stored in the erect image memory 6
A outputs from A to the image rotation circuit 31 and the arrangement in the segment is changed, and then stored in the rotation image memory 6B, whereby the rotation image obtained by rotating the erect image by 90 degrees is held in the rotation image memory 6B. Is done. On the other hand, from the fingerprint database 5, the fingerprint data D1 to be compared is
After being output to the image rotation circuit 31 and the arrangement in the segment is changed, it is stored in the database rotation image memory 5B, thereby generating a rotation image corresponding to the rotation image memory 6B.

Subsequently, the fingerprint data D2 is inclined by a predetermined angle in the horizontal direction (FIGS. 49 and 50) so as to correct the inclination detected by the inclination detection processing, and a line composed of 64-pixel image data is obtained. Eight images are sequentially cut out in the vertical direction, and each linear image data is set in the latch circuit 41 of the matching units 40A to 40B (FIG. 16). Thereby, the inclination of the fingerprint image held in the fingerprint database 5 is corrected, and the fingerprint matching accuracy is improved. The fingerprint data D1 to be compared in the fingerprint database 5 is supplied to the collating units 40A to 40B simultaneously and in parallel so that the fingerprint data D1 is continuous in the raster scanning order.
In the above, it is determined whether or not each bit matches between these two image data.

Further, the number of matched bits is counted by the counter 4
It is counted by 5 and it is judged by the comparing circuit 47 whether or not it exceeds a certain reference value. If the value exceeds the predetermined value, the coordinate data of the corresponding fingerprint data D1 is recorded in the coordinate group memory 49 based on the address data AD of the memory control circuit 30 for the fingerprint database 5.

As a result, the eight linear images are raster-scanned on the image based on the fingerprint data D1 (FIG. 5).
1) The degree of similarity is sequentially detected between the overlapping images, and the coordinate data of a part similar to a certain value or more is stored in the coordinate group memory 49. At this time, the fingerprint collation apparatus 1 scans these linear images with eight collation units 4.
By executing in parallel and simultaneously by 0A to 40H,
The fingerprint collation can be performed in that short time.

Further, when the linear image is cut out and set in the collating units 40A to 40H in this manner, the fingerprint data D2 is set so as not to cut out the linear image for the blackened segment. This effectively avoids a decrease in fingerprint matching accuracy. In addition, a frame for cutting out the linear image is set based on an effective area detected in units of segments in advance, so that only a necessary image can be cut out in a short time. Further, for the fingerprint data cut out in a linear manner, the switching of the logic level is counted, and the portion where the count value is equal to or less than the predetermined value is excluded from the target, so that the portion where the significant information is sufficiently included in the fingerprint collation is included. It is cut out, and the accuracy of fingerprint collation can be improved also by this.

On the other hand, in the fingerprint data D1 of the fingerprint database 5, the operations of the comparison circuits 44 and 47 are switched so as to mask the blackened segments and the blank areas, thereby improving the accuracy of the fingerprint collation. Be improved.

The coordinate data recorded in this manner is:
By the same processing as that at the time of fingerprint registration, the matching units 40A to 40H
, The number of combinations that satisfy the relative positional relationship of the linear images is detected (FIGS. 36 and 37), and the matching rate is detected based on the detected number. At this time, in the system control circuit 3, the matching unit 40
From the partial combinations corresponding to A to 40C, the combinations that do not satisfy the relative positional relationship are excluded from the processing target, and the coordinate values where a high degree of similarity is detected by chance are excluded from the inspection target, The matching ratio is detected based on the remaining combinations. As a result, the collation rate can be detected in a short time, and the time required for the collation can be shortened.

The fingerprint data D thus corrected for tilt is
When the collation rate is detected for No. 2, subsequently, the inclination angle θ is updated (FIG. 53), the same processing is executed, and the frame is switched to execute the same processing (FIG. 54). . Further, when this series of processing is completed for the erect image, the same processing is executed for the rotated image by sequentially updating the angle θ and the frame (FIG. 52). Further, when the processing for the erect image and the rotated image is completed, the same processing is repeated for the remaining fingerprint data D2 fetched into the test fingerprint memory 6 by switching the threshold value (FIG. 47,
Steps SP202 and 203).

When a plurality of collation rates are detected corresponding to the angles and threshold values of the erect image and the rotated image,
In the erect image, it is detected whether or not the collation rate is equal to or more than a predetermined value in all of the switched frames. Subsequently, similarly, in the rotated image, whether or not the collation rate is equal to or more than a predetermined value is detected in all of the switched frames. . Furthermore, in this erect image and rotated image,
It is detected whether the coordinate values detected for the corresponding frame satisfy the relative positional relationship, and similar processing is detected between the switched frames. When all of these conditions are satisfied, it is determined that the fingerprint based on the fingerprint data D2 matches the fingerprint based on the fingerprint data D1. On the other hand, if any one of the conditions is not satisfied, the result of the determination of coincidence is reserved.

As a result, discrimination is made based on the erect image and the rotated image, and the discriminating power is improved. At this time, the discriminating power is further improved by taking into account the relative positional relationship between the coordinate values of the frames as a criterion and further adding conditions to the adjacent frames.

In this determination, it is determined whether or not the collation rates obtained by switching the angle and the reference voltage are the same in the erect image and the rotated image and in each frame by a logical sum. With respect to the identified person, the discrimination power is improved, and the matching result can be reliably output.

Thus, when the determination of the match is reserved (FIG. 41), the reference frame is moved (step SP163), and the same processing is repeated again.
Thereby, the accuracy of fingerprint collation is improved. Even if the frame is further moved, if the matching determination is reserved (FIG. 42), if there is another registered fingerprint data for the same user, the target fingerprint data D1 is switched according to the priority ( Step SP165), and the same processing is repeated, thereby improving the accuracy of fingerprint collation.

Further, even if such a process is repeated, if only a low collation rate equal to or less than a certain value is obtained, it is determined that they do not match, thereby improving the accuracy of fingerprint collation. On the other hand, when the collation rate is constant in a certain range, the criterion is reduced, and the same processing is repeated. Thus, even when the finger is abnormally dry, etc., it is ensured. Identify yourself.

On the other hand, when it is determined that they match, after the matching determination result is output, the matching rate at this time is recorded in the fingerprint database 5 (FIGS. 15 and 41). Subsequently, a change in the past collation rate recorded in the fingerprint database 5 is confirmed, and when the collation rate is reduced, the update of the registered fingerprint data is prompted (FIG. 4).
3). As a result, even when the size of the finger or the like changes, the collation rate is restored, and fingerprint collation can be performed reliably.

(3) Effects of the Embodiment According to the above configuration, the image output from the thinning circuit 20 is divided into equal segments in the horizontal and vertical directions in units of eight pixels to form a plurality of segments. By correcting the reference voltage REF in each segment unit, the binarized signal S1 can be correctly output regardless of whether the fingerprint is partially deformed by the pressing force or the surface state of the isosceles triangular prism 11 changes. It is possible to output a fingerprint, and thus, even with a simple configuration, it is possible to reliably perform fingerprint collation.

Also, since the effective area is detected in units of segments by excluding the segments in which the area where the luminance level rises are equal to or less than a predetermined value, various processes are executed based on the effective areas in units of the segments. Thus, the fingerprint collation processing can be simplified.

That is, by extracting a linear image from the effective area and selecting a registered image, and by executing fingerprint matching processing, unnecessary processing can be omitted and fingerprint matching can be surely performed. it can.

At this time, by setting a frame in the effective area and supplying image data so as to scan the effective area, unnecessary processing can be omitted and the entire processing can be simplified.

Further, by detecting the inclination of the finger from the segment of the effective area and outputting a warning based on the detection result, and correcting the inclination, it is possible to effectively avoid the deterioration of the fingerprint collation accuracy due to the inclination. Can be.

Furthermore, by detecting the placement position of the finger from the number of segments in the effective area and outputting a warning as necessary, the finger can be correctly placed and the fingerprint can be imaged. Accuracy can be improved.

Further, at the time of fingerprint registration, imaging is performed by varying the thinning rate based on the number of segments in the effective area, and at the time of fingerprint collation, imaging is performed at this thinning rate, and the imaging result is large enough for fingerprint collation. This can also improve the accuracy of fingerprint collation. By generating a warning as needed, it is possible to prevent erroneous determination due to abnormal drying of a finger or the like.

Further, by detecting the signal level for each segment of the effective area, and detecting a black-out area in which it is difficult to ensure a sufficient ratio between a bright area and a dark area,
The blackout area is excluded from the processing target, and the fingerprint matching accuracy can be improved.

Further, the blank segment other than the effective area is masked by setting it to a certain logical level, and the criterion in the comparison circuit 47 is changed correspondingly, so that the fingerprint matching accuracy is reduced accordingly. Can be improved.

(4) Other Embodiments In the above-described embodiment, a case has been described in which a plurality of segments are formed at equal intervals in units of eight pixels, but the present invention is not limited to this. The number of pixels to be divided can be freely selected as needed.

In the above-described embodiment, the case where reference voltage REF is corrected in each segment unit has been described. However, the present invention is not limited to this. The reference voltage REF may be corrected in units of a plurality of segments as necessary.

Further, in the above-described embodiment, the case where the threshold value is corrected and the light amount unevenness is corrected has been described. However, the present invention is not limited to this. The unevenness correction processing may be omitted.

Further, in the above-described embodiment, the case where the effective area is detected by excluding the segment where the area where the luminance level rises is equal to or smaller than a predetermined value has been described.
The present invention is not limited to this, by detecting a segment whose area where the luminance level falls is equal to or less than a certain value,
The effective area may be directly detected, or, depending on the optical system or the like, the effective area may be detected by detecting the opposite logic level.

Further, in the above-described embodiment, a case has been described in which a frame is set in an effective area and a linear image is cut out. However, the present invention is not limited to this. May be cut out.

Also, in the above-described embodiment, the case where a linear image is cut out from the effective area and the fingerprint is collated has been described. However, the present invention is not limited to this. You may collate.

Further, in the above-described embodiment, a case has been described where the inclination of the finger is detected from the segment of the effective area and a warning is output if the inclination is large. However, the present invention is not limited to this, and the inclination is simply corrected. Fingerprint collation may be performed simply by doing.

Further, in the above-described embodiment, the case where the placement position of the finger is detected from the number of segments of the effective area has been described. However, the present invention is not limited to this, and the finger is detected based on the coordinate data of the effective area. May be detected.

Further, in the above-described embodiment, a case has been described in which a warning is output when the placement position of the finger is determined to be abnormal. However, the present invention is not limited to this. Alternatively, the displacement of the mounting position may be automatically corrected.

In the above-described embodiment, the case where the thinning rate is varied based on the number of segments in the effective area has been described. However, the present invention is not limited to this. Signal D
The resolution of V1 may be switched.

Also, in the above-described embodiment, a case has been described in which the image to be inspected is cut out linearly and fingerprint collation is performed. However, the present invention is not limited to this, and the fingerprint image side registered in the fingerprint database is linearly formed. The clipping may be performed for fingerprint matching, and the present invention can be widely applied to fingerprint matching based on feature points.

Further, in the above-described embodiment, the case where the present invention is applied to the fingerprint collating apparatus has been described. However, the present invention is not limited to this. The present invention can be widely applied to an image collating device that determines similarity or ratio similarity in whole or in part.

[0276]

As described above, according to the present invention, a binarized image obtained by binarizing a video signal is divided into meshes to form a plurality of segments. By performing the processing of (1), it is possible to surely perform image matching with a simple configuration.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of a fingerprint matching device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a fingerprint data input unit of the fingerprint matching device of FIG.

FIG. 3 is a schematic diagram illustrating a fingerprint image by a fingerprint data input unit in FIG. 2;

FIG. 4 is a signal waveform diagram for explaining correction of light amount unevenness in the fingerprint data input unit in FIG. 2;

FIG. 5 is a plan view showing a relationship between a fingerprint image and a segment.

FIG. 6 is a signal waveform diagram for explaining actual correction of light amount unevenness;

FIG. 7 is a signal waveform diagram for explaining correction of a threshold value composed of a reference voltage.

FIG. 8 is a signal waveform diagram for explaining switching of a threshold value including a reference voltage.

FIG. 9 is a block diagram showing a test fingerprint memory and a fingerprint database together with peripheral circuits.

FIG. 10 is a schematic diagram illustrating a fingerprint image by the fingerprint data input unit in FIG. 2;

11 is a schematic diagram illustrating an image obtained by rotating the fingerprint image of FIG. 9;

FIG. 12 is a block diagram illustrating the image rotation circuit of FIG. 9;

FIG. 13 is a schematic diagram for explaining image rotation by the image rotation circuit of FIG. 12;

FIG. 14 is a schematic diagram illustrating an image after rotation corresponding to FIG. 13;

FIG. 15 is a schematic diagram illustrating the contents of a fingerprint database.

FIG. 16 is a block diagram illustrating a matching rate detection unit.

FIG. 17 is a schematic diagram for explaining the extraction of a linear image.

FIG. 18 is a schematic diagram illustrating supply of fingerprint data to a linear image.

FIG. 19 is a schematic diagram illustrating the contents of a coordinate group memory.

FIG. 20 is a flowchart illustrating a processing procedure of the system control circuit 3.

FIG. 21 is a flowchart showing an optical system operation confirmation process in FIG. 20;

FIG. 22 is a flowchart showing the fingerprint registration processing of FIG.

FIG. 23 is a flowchart showing the actual registration processing of FIG. 22.

FIG. 24 is a flowchart showing a continuation of FIG. 23;

FIG. 25 is a flowchart showing an effective area detection process in FIG. 23;

FIG. 26 is a schematic diagram used for explaining the effective area detection processing in FIG. 25;

FIG. 27 is a flowchart showing a finger position determination process of FIG.

FIG. 28 is a flowchart illustrating a zoom process of FIG. 23;

FIG. 29 is a flowchart illustrating a threshold value correction process of FIG. 23;

FIG. 30 is a flowchart showing the blackout processing of FIG. 23;

FIG. 31 is a flowchart showing an angle detection process of FIG. 23;

FIG. 32 is a schematic diagram used for describing the angle detection processing in FIG. 31.

FIG. 33 is a schematic diagram used to explain a process following FIG. 32;

FIG. 34 is a flowchart showing a registration data selection process of FIG. 24;

FIG. 35 is a flowchart showing a correlation value detection process of FIG. 34;

FIG. 36 is a flowchart showing a continuation of FIG. 35;

FIG. 37 is a flowchart showing a continuation of FIG. 36;

FIG. 38 is a flowchart showing a continuation of FIG. 37.

FIG. 39 is a schematic diagram for explaining the extraction of a linear image from a rotated image.

FIG. 40 is a schematic diagram illustrating supply of fingerprint data to a linear image in a rotated image.

FIG. 41 is a flowchart showing the fingerprint registration processing of FIG. 20.

FIG. 42 is a flowchart showing a continuation of FIG. 41.

FIG. 43 is a characteristic curve diagram for explaining a decrease in a matching rate.

FIG. 44 is a schematic diagram explaining movement of a frame.

FIG. 45 is a flowchart showing the fingerprint data input processing of FIG. 41.

FIG. 46 is a flowchart showing a zoom process of FIG. 45.

FIG. 47 is a flowchart showing a comparison determination process in FIG. 41.

FIG. 48 is a flowchart showing a continuation of FIG. 47.

FIG. 49 is a schematic diagram for explaining tilt correction;

FIG. 50 is a schematic diagram for explaining processing of image data in the inclination correction of FIG. 49;

FIG. 51 is a schematic diagram illustrating supply of fingerprint data to a linear image whose inclination has been corrected;

FIG. 52 is a schematic diagram illustrating the case of the rotated image of FIG. 51;

FIG. 53 is a schematic diagram for explaining the switching of angles.

FIG. 54 is a schematic diagram used to explain the switching of frames.

FIG. 55 is a schematic diagram for explaining coordinate values in an erect image.

FIG. 56 is a schematic diagram illustrating coordinate values in a rotated image in relation to the erect image in FIG. 55;

FIG. 57 is a schematic diagram showing a relationship between coordinate values detected from each frame.

[Explanation of symbols]

1 ... Fingerprint collation device, 3 ... System control circuit, 4 ...
Fingerprint data input unit, 5 ... Fingerprint database, 6 ... Test fingerprint memory, 6A ... Erect image memory, 6B ... Rotation image memory, 7 ... Collation rate detection unit, 7B ... Rotation image memory for database, 11 isosceles triangular prism,
14 comparison circuit, 16 light intensity correction memory, 20
Thinning circuit 22, pulse wave detector 23, pressure sensor 24, threshold value correction memory 30, memory control circuit 31, image rotation circuit 40A to 40H collating unit 49 … Coordinate group memory

Claims (13)

[Claims] An image signal is binarized by a predetermined threshold value to generate a binarized signal, and based on the binarized signal,
An image matching device that determines whether the image matches or does not match between a binarized image based on the binarized signal and a predetermined reference image, wherein the binarized image is divided into meshes to form a plurality of segments. And generating the binarized signal by setting the threshold value for each segment. 2. A binarized signal is generated by binarizing a video signal according to a predetermined threshold value, and based on the binarized signal,
An image matching device that determines whether the image matches or does not match between a binarized image based on the binarized signal and a predetermined reference image, wherein the binarized image is divided into meshes to form a plurality of segments. An image matching device for detecting a signal level of the binarized signal for each segment and detecting an effective area obtained by imaging a matching target from the binarized image. 3. The image matching apparatus according to claim 2, wherein the image matching is determined by comparing the binarized image with the reference image in the effective area. . 4. A method for extracting a plurality of linear images from within the effective area, sequentially displacing each of the linear images on the reference image, and obtaining a comparison result between overlapping pixels to obtain the line image. A coordinate value having a high correlation between a linear image and the reference image is sequentially detected, and, between the linear images, a relative position relationship between the coordinate values having a high correlation is used as a reference, and the image coincidence or 3. The image matching apparatus according to claim 2, wherein a mismatch is determined. 5. The image collating apparatus according to claim 2, wherein the inclination of the collation target is detected based on the coordinate data of the effective area. 6. A method for determining a position of the matching target in the binarized image based on the number of segments of the effective region or coordinate data of the effective region, or determining the position of the matching target in the binarized image. The image collating apparatus according to claim 2, wherein the position of the image is determined and a determination result is output. 7. The image matching method according to claim 2, wherein the number of segments in the effective area is counted, and the size of the matching target in the binarized image is detected based on the counting result. apparatus. 8. The image matching apparatus according to claim 7, wherein a warning is issued when the size of the matching target is equal to or smaller than a predetermined value. 9. The image matching apparatus according to claim 7, wherein when the size of the object to be checked is equal to or smaller than a predetermined value, the magnification of the image based on the binarized signal is switched. 10. The image matching apparatus according to claim 2, wherein a signal level of the binarized signal is set to a constant value for segments other than the effective area. 11. A signal level of the binarized signal is detected for each segment of each of the effective areas, and a ratio between a light portion and a dark portion is secured to a predetermined value or more based on a detection result of the signal level. The image matching apparatus according to claim 2, wherein a segment that is difficult to detect is detected. 12. The method according to claim 1, wherein the number of segments in which it is difficult to secure the ratio between the light portion and the dark portion equal to or more than a predetermined value is counted, and a warning is output when the count value is equal to or more than the predetermined value. 12. The image matching device according to item 11. 13. A binarized signal is generated by binarizing a video signal according to a predetermined threshold value, and based on the binarized signal, a binarized image based on the binarized signal and a predetermined reference An image collating apparatus that determines whether an image matches or mismatches with an image, wherein the binarized image is divided into meshes to form a plurality of segments, and the 2 in a predetermined segment in the plurality of segments is formed. An image matching apparatus, comprising: detecting a signal level of a digitized signal to detect the presence / absence of a matching target in the binary image.