JP2613959B2 - Fingerprint pattern classification device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fingerprint pattern classification apparatus, and more particularly to a fingerprint pattern classification apparatus for automatically classifying patterns of collected fingerprints.

2. Description of the Related Art Conventional classification methods using features obtained from ridge directions include matching with a previously created standard pattern (Goto, Nakamura, Minami, Okuno; “Fingerprint image classification by direction distribution pattern”, IEICE Technical Report vol.81, no.210, IE81-88)
Or, the ridge is regarded as a vector field, and the values of the integral of the rotation and the area of the divergence are used (Noda, Oteru, Kobayashi, Kato;
“Fingerprint pattern classifier”, IEICE, PRL73-97) A delta (hereinafter, called delta) that is formed in a triangular shape by the center (hereinafter, called core) and ridges of a fingerprint that is a reference of classification. A method of classifying according to presence / absence, number, and position has been proposed.

In the above-described conventional method, the presence or absence of a delta plays an important role in the classification process, but has a drawback that it is difficult to collect the delta depending on the location on the fingerprint. In particular, a fingerprint collation device for criminal investigations targeting latent fingerprints often processes fingerprints lacking deltas, and cannot be used for delta classification.

Further, the extraction of cores and deltas by standard pattern matching has a problem in that the extraction of the core and delta has a great effect on the extraction due to a standard pattern created in advance.

SUMMARY OF THE INVENTION Therefore, the present invention has been made to solve the drawbacks of the related art, and the object of the present invention is to enable stable and accurate pattern classification without resorting to unstable delta classification. To provide a fingerprint pattern classifying apparatus.

The fingerprint pattern classifying apparatus according to the present invention comprises a means for collecting a fingerprint image and outputting it as two-dimensional image data, dividing the two-dimensional image into a plurality of regions in a grid pattern, and Means for detecting the ridge direction of the fingerprint in the area, means for detecting the center position of the fingerprint from each ridge direction of the area, and detecting the shape of the ridge from the ridge direction and the center position Means, and means for determining the tendency of the ridge direction according to the most ridge direction among the ridge directions of the region, and for each of a plurality of fingerprint classes predetermined by the shape of the ridge and the tendency of the ridge direction. Classification means for classifying fingerprint patterns;
Means for determining the quality of the fingerprint image based on whether the number of areas in which the ridge direction has been detected and determined is equal to or greater than a predetermined number, and means for correcting the image by a predetermined correction method when the quality is determined to be low quality And the fingerprint classification is performed again based on the corrected image.

Embodiment Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.
In the present embodiment, a fingerprint image placed on a transparent body or a fingerprint image imprinted on an original printing paper is collected, an image input unit 10 for photoelectric conversion, and a photoelectric conversion signal is converted into two-dimensional quantized image data. A / D conversion unit 11 for conversion, an image input control unit 12 for performing input control of a fingerprint image, an image storage unit 13 for storing two-dimensional quantized image data, and information on a fingerprint image to be classified are input. An input unit 19, an output unit 20 that outputs a classification result and various messages, and a classification processing unit 18 that performs classification by inputting image data from the image storage unit 13.
And

The classification processing unit 18 receives the image data from the image storage unit 13 and the information on the fingerprint image to be classified from the input unit 19 as input and outputs the external interface 14 to the output unit 20.
, A working memory 17, a program memory 16 storing a program, and a central processing unit 15 controlled by the program.

The fingerprint input unit 10 is described in JP-A-54-69300 and JP-A-54-85600, and describes an optical boundary condition by light from a light source for a finger placed on a transparent body. A device for inputting a photoelectric conversion image of a fingerprint pattern using an imaging device such as an ITV (Industrial television) or a fingerprint image on an original printing paper described in Japanese Patent Application Laid-Open No. 55-138174. Devices are available and can be used.

In the present invention, the ridge shape (middle in FIG. 2) is detected for the upper tier pattern as shown in FIG. 2, and the tendency of the ridge direction (lower tier in FIG. 2) is determined. Perform classification. The broken line in the pattern at the top of FIG. 2 indicates the row where the core exists.

FIG. 3 is a flowchart showing the operation of one embodiment of the present invention. First, in step 310, the image storage unit 1
A fingerprint image is input from 3 through the external interface 14. In step 311, features are extracted from the ridge direction, and in step 312, a region including the core is extracted.

FIG. 4 is a diagram showing a processing method of the collected fingerprint image in the present embodiment, and the collected two-dimensional image 40 is formed in a substantially lattice shape.
It is divided into a plurality of areas (hereinafter referred to as zones), and the ridge direction is detected for each zone and processed as zone data 42.

As shown in FIG. 5, the ridge direction for each zone is, for example,
Divide 360 degrees into 16 equal parts and label them in one of 16 directions (0-15)
The zone data, which is the characteristic obtained by the above, is used. In addition, 8
Equal, 32 and 64 equal parts are also conceivable.

In FIG. 4, zone 41 is a region including the core,
Zone 42 is a direction-determined area, zone 43 is an uncertain area in which ridges exist but the direction cannot be determined, and zone 44 represents a non-data area in which ridges do not exist.

In the process of extracting the features (zone data) from the ridge direction in step 311, it is necessary to extract the zone data shown in FIG. 4, and in step 312, the process of extracting the zone 41 including the core is required. However, these processes are disclosed in Japanese Patent Application Laid-Open No. 55-138174 and Japanese Patent Application
These are described in detail in the specifications of Japanese Patent Application No. 099334, Japanese Patent Application No. 63-099335, and Japanese Patent Application No. 63-099336, and will not be described here.

In the next step 313, detection of the shape of the ridge is determined from the zone data obtained in steps 311 and 312 and the zone including the core. The method will be described with reference to FIG.

With the zone 41 including the core as a starting point, in the frame 52, a main scanning direction is a y-axis (an axis perpendicular to the finger joint) from a negative direction to a positive direction, and an x-axis (centering around the zone 41 including the core). By sub-scanning in both positive and negative directions (direction axes parallel to the finger joints), the presence or absence and type of the direction label shown in FIG. This gives y
The shape of the ridge on the axial main scanning line is detected.

For example, in the example of FIG. 6, the shape of the ridge corresponding to the right hoof-shaped pattern (No. 2) shown in the middle part of FIG. 2 is detected.

The method of determining the frame 52 in FIG. 6 is as follows. The upper limit of the frame 52 is a line including the zone 41 including the core,
With this line as a starting point, the main scanning direction is a direction from negative to positive on the x-axis, and sub-scanning is performed from negative to positive on the y-axis. The horizontal tendency H is calculated by the equation.

H = [1.00 × a (0) + 0.05 × {a (1) + a (7)} + 0.05 × {a (2) + a (6)}] / [1.00 × ｛a
(0) + a (4)} + 0.05 × {a (1) + a (3) + a (5) + a (7)} + 2 × 0.05 × {a (2) + a (6)}] where a (n ) Is the number of direction determination zones of the label n existing on the scan.

Then, the value H calculated as described above for each scan is compared with a predetermined threshold value S1, and the first scan line satisfying the condition of H <S1 is set as the lower limit of the frame 52.

When the number of sub-scans is an odd number, the lower limit is one line lower in the negative / positive direction of the y-axis.

The left and right limits of the frame 52 are the left and right limits of the two-dimensional image data 40 as shown in FIG. 6, but the left and right two or three columns are unclear and often cannot be zone data. May be omitted.

The method of determining the ridge direction tendency in step 314 will be described with reference to FIGS. First, a rectangular frame 61 as shown in FIG. 7 is provided, and only the direction fixing zone 42 existing in this frame is targeted. The number of each direction label of the direction fixing zone 42 is calculated, and the height is calculated based on the value. Calculate the tendency in the line direction.

For example, the tendency L in the direction from the lower left to the lower right in the frame 61 is calculated using the following equation.

L = [1.00 × a (2) + 0.05 × {a (1) + a (3)} + 0.05 × {a (0) + a (4)}] / [1.00 × ｛a
(2) + a (6)｝ + 0.50 × {a (1) + a (3) + a (5) + a (7)} + 0.05 × (a (0) + a (4)｝ × 2] By comparing the calculated value L with a predetermined threshold value, the tendency in the ridge direction is determined.

Note that the upper and lower limits of the frame 61 are the same as the upper and lower limits of the frame 52, the left limit is the column including the zone 41 including the core, and the right limit is the next column in the positive direction of the x-axis with respect to this column. is there.

Next, two rectangular frames 62 and 63 as shown in FIG. 8 are provided, and the same calculation as above is performed respectively to determine the tendency of the ridge direction in each frame. In the present example, the tendency in the ridge direction corresponding to the right-flowing hoof pattern (No. 2) shown in the lower part of FIG. 2 is determined.

The determination of the frames 62 and 63 in FIG. 8 is as follows. Frame 6
The lower limit of 2 and the upper limit of the frame 63 are determined by equally dividing the width from the upper limit to the lower limit of the frame 52 in FIG. The upper limit of the frame 62 is the upper limit of the frame 52, and the lower limit of the frame 63 is the lower limit of the frame 52.

The main scanning is performed from the negative to the positive direction of the y-axis, and the sub-scanning is performed in the positive and negative directions of the x-axis centering on the zone 41 including the core. Is calculated, and the vertical tendency V is determined by the following equation.

V = [1.00 × a (4) + 0.50 × {a (3) + a (5)} + 0.05 × {a (2) + a (6)}] / [1.00 × ｛a
(0) + a (4)｝ + 0.50 × {a (1) + a (3) + a (5) + a (7)} + 2 × 0.05 × (a (2) + a (6)}] The value V calculated by the equation is compared with a predetermined threshold value S2, and the first row that satisfies the condition of V <S2 is set as the leftmost and rightmost columns of the frames 62 and 63.

The reason why the tendency in the ridge direction is determined in step 314 in two steps using the method shown in FIGS. 7 and 8 is as follows.

That is, first, it is determined whether the tendency in the ridge direction obtained using the frames 62 and 63 corresponds to the ordinary bow-shaped pattern (No. 5) shown in the lower part of FIG. The final ridge direction tendency is determined by determining which one of No. 1 to No. 4 in the lower part of FIG.

In step 315, classification is performed by associating the shape of the ridge detected in step 313 with the tendency of the ridge direction determined in step 314 in FIG.

In step 316, it is determined using the uncertainty zone 43 whether or not the input fingerprint image is of low quality. This determination method will be described with reference to FIG. First, step 315
For the input fingerprint image classified other than the ordinary bow-shaped pattern (No. 5) in the classification in, a frame 52 is provided as shown in FIG. , And scans from the negative to the positive direction of the y-axis starting from the zone 41 including the core.

Here, it is assumed that only the uncertainty zone 43 is the target, and the input fingerprint image classified as a right-foot-shooting pattern (No. 2) in step 315 has a positive x-coordinate centered on the zone 41 including the core. The direction is defined as sub-scanning. For the input fingerprint image classified as the left footprint (No. 3) in step 315, the negative direction of the x coordinate is defined as sub-scanning with the zone 41 including the core as the center. With respect to the input fingerprint image classified as the protruding bow-shaped pattern (No. 4), the positive and negative x-coordinates of the input fingerprint image in the y-axis direction main scanning are set to the sub-scanning around the zone 41 including the core. The rate at which the defined zone 43 exists is calculated, and it is determined whether or not the input fingerprint image is of low quality based on the number of main scans in which the uncertain zone 43 is present at a certain rate or more (hereinafter, referred to as a defective column). .

Further, regarding the input fingerprint image classified as a normal bow-shaped pattern (No. 5) in step 315, the uncertain zone 43 existing in each of the two rectangular frames 62 and 63 shown in FIG. It is determined whether the quality is low based on the ratio.

In step 317, it is determined whether the low-quality input fingerprint image determined in step 316 is a target of classification rejection (hereinafter, referred to as rejection) or a target of reclassification. In this method, the determination is made based on the number of defective columns used in step 316. The ordinary bow-shaped print determined as the input fingerprint image determined to be of low quality in step 316 is to be rejected. For the input fingerprint image determined to be rejected in this step 317, the output unit 20 outputs the result of this step 317 through the external interface 14 in step 318.

In step 319, the input fingerprint image determined to be a target of the re-classification in step 316 is corrected. The method is performed by using a well-known established relaxation method, and after correction, step 312 and subsequent steps are performed again. In addition, about this established relaxation method, "Rasenfeld: ITERTIVE METHODS I
N IMAGE ANALYSIS ", Proc.IEEE Conf.P.R & I.P14 / 18 (19
77).

In step 320, the shape and position of the ridge detected in step 313 and the tendency of the ridge direction determined in step 314 are associated with FIG. It is determined whether or not there is a second classification candidate.

If it is determined in step 320 that the input fingerprint image to be classified does not have the second classification candidate, in step 318 the output unit 2
From 0, the classification result of step 315 is output.

In step 321, a second classification candidate of the input fingerprint image to be subjected to the classification processing determined to have the second classification candidate is determined. The method is based on the classification result of the input fingerprint image to be classified in step 315 (step 1 in FIG. 9), the ridge shape and its position detected in step 313 (step 3 in FIG. 9), and step 314. The second classification candidate is determined by associating the tendency in the ridge direction (step 4 in FIG. 9) determined in step (2) with the step in FIG. 9 (step 2). The arrow in the third row of FIG. 9 indicates the distance between the detected ridge shape and the row including the core.

If it is determined in step 320 that the input fingerprint image to be classified has the second classification candidate, in step 318 the output unit 20 is output through the external interface 14.
Thus, the classification result in step 315 and the second classification candidate determined in step 321 are output.

As described above, in the present embodiment, classification is performed from the position of the core of the zone data.

Effects of the Invention As described above, according to the present invention, according to the zone data used in the existing automatic fingerprint matching system, to perform classification by detecting the shape of the ridge and determining the tendency of the ridge direction, There is an effect that stable fingerprint pattern classification can be performed. Furthermore, the classification effect can be obtained also for the classification of the latent fingerprint in which the existence of the delta is not guaranteed.

[Brief description of the drawings]

FIG. 1 is a system block diagram of an embodiment of the present invention, FIG. 2 is a diagram showing the concept of a pattern classification method used in this embodiment, and FIG.
FIG. 4 is a flowchart showing the operation of the embodiment of the present invention.
FIG. 5 is a diagram showing an example of zone data of a collected fingerprint image, FIG. 5 is a diagram showing direction labels of ridges in each zone, FIG. 6 is a diagram for explaining the detection of ridge shapes, FIG. 7 and FIG. FIG. 8 is a diagram for explaining the calculation of the tendency in the ridge direction, and FIG. 9 is a diagram showing the concept of the second classification candidate. Explanation of reference numerals of main parts 10 fingerprint image input unit 13 image storage unit 40 two-dimensional image data 15 central processing unit 41 core zone 17 working memory 42 Direction determination zone 18 Classification processing unit 43 Undefined zone 44 Non-data zone

Claims (3)

(57) [Claims] 1. A means for collecting a fingerprint image and outputting it as two-dimensional image data, dividing the two-dimensional image into a plurality of areas in a grid pattern, and for each of these areas, the height of the fingerprint in the area is determined. Means for detecting a line direction, means for detecting the center position of the fingerprint from each ridge direction of the region, means for detecting the shape of the ridge from the ridge direction and the center position, and Means for determining a tendency of a ridge direction according to the most ridge direction, and a classification for classifying fingerprint patterns for each of a plurality of fingerprint classes predetermined according to the shape of the ridge and the tendency of the ridge direction. Means, means for determining the quality of the fingerprint image based on whether the number of areas in which the ridge direction has been detected and determined is equal to or greater than a predetermined number, and when the quality determination determines that the quality of the fingerprint image is low, the image is determined by a predetermined correction method. Means for correcting Fingerprint pattern classification device is characterized in that so as to form a re fingerprint classification by the positive image after. 2. The fingerprint pattern classification according to claim 1, further comprising means for classifying up to a second candidate when the similarity is high and similar to a plurality of fingerprint classifications at the time of classification by said classification means. apparatus. 3. A means for collecting a fingerprint image and outputting it as two-dimensional image data, dividing the two-dimensional image into a plurality of regions in a grid pattern, and for each of these regions, the height of the fingerprint in the region is determined. Means for detecting a line direction, means for detecting the center position of the fingerprint from each ridge direction of the region, means for detecting the shape of the ridge from the ridge direction and the center position, and Means for determining a tendency of a ridge direction according to the most ridge direction, and a classification for classifying fingerprint patterns for each of a plurality of fingerprint classes predetermined according to the shape of the ridge and the tendency of the ridge direction. Means, at the time of classification by said classification means, when the similarity is high and similar to a plurality of fingerprint classifications,
A fingerprint pattern classification device comprising means for classifying up to a second candidate.