JPH0782541B2 - Fingerprint data flow controller based on fingerprint quality - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a feature point extraction process and a feature point confirmation process in a pattern recognition system for automatically matching a fingerprint based on a fingerprint feature point, and more particularly to a fingerprint image. The present invention relates to a control device that controls the flow of fingerprint image data by using quality information to confirm and correct the fingerprint image data.

[Conventional technology]

Conventionally, in pattern recognition systems that perform automatic fingerprint collation, the fingerprint is read using a CCD scanner, image data with shades is obtained, and feature points are automatically extracted from this image data. Therefore, the feature extraction rate from the fingerprint is 100%.
% Is generally impossible.

For this reason, the extraction contents of feature points are confirmed by a man machine (CRT, etc.), and as a result, the extraction contents are corrected if necessary. Generally, fingerprint data is displayed on the display for this confirmation and correction. In this case, all the image information read from the fingerprint reading device is simply sent to the fingerprint monitoring device and displayed on the display. Then, the confirmation / correction work of the fingerprint data is entirely executed by the operator's judgment.

[Problems to be solved by the invention]

By the way, a fingerprint reader and a plurality of fingerprint monitor devices are provided, and the writing of the fingerprint image data obtained by the fingerprint reader into the original image file and the display from the original image file to the plurality of fingerprint monitor devices are executed asynchronously. In the case of a system provided with a fingerprint monitor device in consideration of the balance of the processing time of the writing process and the display process, if all the fingerprint image information obtained from the fingerprint reading device is displayed on the fingerprint monitor device, much time is required for data transfer. In addition, there is a problem that the capacity of a file such as a disk for temporarily storing the fingerprint image data becomes enormous.

On the other hand, as described above, in the pattern recognition system that performs automatic fingerprint collation, it is necessary to confirm and correct the fingerprint image data in order to improve the collation accuracy. Therefore, when all images are displayed on the fingerprint monitor and the confirmation and correction of the fingerprint image information is performed, the operator must determine whether or not to correct the fingerprint image information. There is a problem that the quality of the fingerprint image information becomes non-uniform due to variations in the correction content.

[Means for solving problems]

The fingerprint data flow controller according to the fingerprint quality of the present invention is
A fingerprint reading means for extracting a feature point list including fingerprint quality information from a fingerprint image, a core line, and an original image; a fingerprint quality determining means for determining whether to input the core line or the original image based on the fingerprint quality information; The feature point list is stored in file means, and the output means for outputting the core line or the original image to the file means according to the determination of the fingerprint quality determination means, the image search means for searching the file means, and the image search means And a fingerprint monitor unit for displaying the retrieved search contents and for confirming and correcting the retrieved contents, and according to the fingerprint quality information,
It is characterized in that the display is unnecessary or the feature point list and the core line are displayed, or the feature point list, the core line and the original image are displayed.

〔Example〕

 The present invention will be described below with reference to examples.

Referring to FIG. 1, a fingerprint reading device 2, a plurality of fingerprint monitor devices 3, and a magnetic disk device 4 are connected to a central processing unit (CPU). The central processing unit 1 includes an image input unit 1a, a fingerprint quality determination unit 1b, an image output unit 1c, and an image search unit 1
equipped with d. The fingerprint reading device 2 also includes a fingerprint quality extraction unit 2a.

The fingerprint reader 2 is shown in FIG. 3 (a), for example, from a fingerprint base paper in which fingerprints for 10 fingers per person are pressed with ink.
As shown in, the position coordinates M (x, y) of the fingerprint feature points (end points and branch points), the ridge direction d of the fingerprint, and the neighboring feature points (R extracting ₁ to r ₄₎ ridge count between the (r ₁ ~r _4). Then, a reading number (processing number) is printed on the back surface of the fingerprint base paper read by the fingerprint reading device 2.

The feature points extracted by the fingerprint reader 2 and the number of ridges between neighboring feature points are used during fingerprint matching, and the number of ridges (relation) between neighboring feature points is compared as shown in FIG. 3 (b). Used as important information. In the example shown in FIG. 3 (b), the number of ridges between the feature points M ₁ and M _{2 is} the same for three fingerprints (1) and (3),
It is shown that the fingerprint (2) is one and different from the fingerprint (1).

Fingerprint information (fingerprint data) output from the fingerprint reader 2
Is a feature point list, a core line, and an original image as shown in FIG. As shown in FIG. 4B, the feature point list includes a recognition section [I (Identification) -Part], a zone section [Z (Zone) -Part], and a fine section [M (Minuti).
ae) -Part], and I-Part contains the reading number of fingerprint data (personal descriptive information such as gender and date of birth), coordinate axes (X _c , Y _c , D _c ), and fingerprint quality. In Z-Part, each area (0.8 mm × 32 mm in the fingerprint image (32 × 40 mesh))
0.8mm) is clear or unclear information is expressed by 2 bits as shown in Fig. 4 (c), 00: unclear area, 01: unstable area, 10: clear area, 11: curved part To be done. As mentioned above, information on whether each area is clear or unclear (clear and unclear information)
Is represented by 2 bits. The blurred area is represented by "00", the unstable area is represented by "01", the blurred area is represented by "10", and the curved portion is represented by "11". Generally, in a fingerprint base paper in which fingerprints are pressed with ink, there may be a portion where no clear ridge is observed. In other words, there is a portion where the base paper (base) and the ridge cannot be distinguished (for example, a portion where the shade is unclear due to ink coverage). In this way, a portion where the base paper (base) and the ridge cannot be distinguished is defined as the unclear area. Further, a portion where the distinction between the base paper (base) and the ridge can be clearly identified is defined as a clear area. Furthermore, in the fingerprint base paper,
For example, there are parts such as the center of the fingerprint and the triangle where the shade is clear but the direction of the ridge is not constant. Such an area is a curved portion. Then, the unclear area, the clear area, and the area other than the curved portion are set as unstable areas. That is, a region that cannot be determined to be a blurred region, a clear region, or a curved portion is an unstable region. Furthermore, M
−Part includes feature points (position coordinates x, y, ridge direction d, neighboring feature points
R _{1 to} R ₄ and the number of ridges between neighboring feature points r _{1 to} r ₄ ) are set.

The core line is obtained by extracting the skeleton of the fingerprint ridge as shown in FIG. 4 (d), and 512 × 6 as shown in FIG. 4 (e).
Each area of 40 mesh (0.05mm x 0.05mm) is represented by 1 bit (0: not core line, 1: core line). On the other hand, the original image is on the fingerprint original paper as shown in Fig. 4 (f). The ink density of the fingerprint ridges pressed against the
The density of each area (0.05 mm × 0.05 mm) of 640 mesh is expressed by 4-bit, 16-level gradation (FIG. 4 (g)).

The fingerprint monitor device 3 has feature points, coordinate axes, zones (ZON
E), core line, original image is displayed on the screen, and it is used as a man-machine device for correcting coordinate axes, zones, and core lines. The operator corrects the coordinate axis, ZONE, and the core line with a stylus pen (not shown) of the tablet unit of the fingerprint monitor device 3, and inputs personal descriptive information and commands from the keyboard unit. Further, a hard copy of the image can be taken from the fingerprint monitor device 3 if necessary.

The magnetic disk device 4 is a file that temporarily stores an original image file, and a feature point list, a core line, and an original image are temporarily stored in the original image file.

Here, the fingerprint quality extracted from the original fingerprint image by the fingerprint quality extraction unit 2a of the fingerprint reading device 2 will be described.

Fingerprint quality is defined by area quality and coordinate axis quality. The area quality expresses the sharpness of the fingerprint impression, and the coordinate axis quality expresses the extraction accuracy of the coordinate axis.

(1) Area quality Area quality is a ratio of the number of clear areas to excellent, good, good, and bad.
It is evaluated in stages. The ratio of the number of clear areas is determined for each section. Referring to FIGS. 5 (a), (b), and (c), the impression center information (I _c , J _c ) and impression edge information (I _l , I _r , J) are used when determining the section. _t , J _b ) and
The partition determination function shown in FIG. 7B is referred to (this partition determination function is determined in advance by experiments using a plurality of fingerprint base papers). At this time, first, Δl = | I _c −I _l | / 16, Δr =
| I _c −I _r | / 16, Δt = | J _c −J _t | / 16, and Δb = | J _c −J _b | / 1
Obtain 6 and read the value on the partition determination function based on these values to determine the partitions A, B, and C. In other words, A _l, B _l from Δl,
And C _l is Motomari, [Delta] r in the same manner, Delta] t, and each of _{Δb A r, B r, C} r, A t, B t, C t and A _{b, B b,} the C _b obtained. Therefore, section A is (A _l , A _r , A _t , A _b ), and section B is (B _l , B _r , B _t ,
B _b), C compartment _{(C l, C r, C} t, is defined by C _b) (FIG. 5 (c)).

When the sections are determined as described above, the quality within each section is then obtained. In-compartment quality A _Z , B _Z , and C _Z are Z-Part
It is determined by the ratio of the number of clear areas of each area in
Here, the following points are taken into consideration.

There must be a certain value or more in the clear area in Section A.

There should not be many white areas in Section B.

There are many white areas in the C section.

Then, the area quality is defined by the combination of intra-compartment qualities A _Z , B _Z , and C _Z. As is clear from the above description,
When setting the partition determination function, a plurality of fingerprint base papers are prepared in advance as samples, and using these samples, a function that represents a partition in which the clear area is a certain value or more is A partition determination function, and a white background area is small. Is set as a B section determination function, and a function representing a section having many white background areas is set as a C section determination function.

(2) Coordinate axis quality Extraction of the coordinate axis consists of extraction of the center point position and extraction in the positive direction of the Y axis.
It is evaluated in 3 grades of "OK" and "NO". FIG. 6 (a),
The center point extraction will be described with reference to (b) and (c). First, the start point (S _{1 in} Figure 6 (a))
~S indicated by _3, but will be described here S _1. ), Trace downward and find the intersection point r ₁ with the ridge. Then, the ridge line corresponding to the intersection point r ₁ is traced in the left-right direction to find the highest position H ₁ . Next, trace downward from the highest position H ₁ to find the intersection r ₂ with the ridge, and similarly, find the highest position H ₂ , the intersection r ₃ , and the highest position H ₃ . If the highest position H ₃ exists on the innermost ridge, the highest position H ₃
Trace downward from to find the intersection point r ₄ with the ridge.
That is, H ₃ → r ₄ is repeated. Then, the highest position H ₃ is defined as the center point.

When the center point is defined, the extraction process in the Y-axis positive direction starts.

From the center point obtained as described above, the ridge where the center point exists is traced equidistantly from the center point to the left and right,
Find points l ₁ and r ₁ as shown in FIG. 6 (b). Points l ₁ and r ₁
Find the intersection points l ₂ and r ₂ , and l ₃ and r ₃ of the straight line connecting the two and the two outer ridges. Equidistant in the vertical direction for each ridge based on these intersections (l _{1 to} l ₃ , r _{1 to} r ₃ ) and trace the ridges to find the start and end points of the vector.
L _{1 to} L ₃ and R _{1 to} R ₃ are determined. The length of the direction vector is made shorter as it goes outside the ridge where the center point exists. As shown in FIG. 6 (c), the direction of the sum of the direction vectors is the Y-axis positive direction, which is represented as the Y-axis unit vector in FIG. 6 (c). Coordinate axis quality is determined according to the following criteria in the above-mentioned center point and coordinate axis extraction process.

Coordinate axis quality “Excellent” (a) In extracting the center point position, all the center positions must be found to be the same position as a result of tracing from three start positions.

(A) The center point position should be found near the center of the fingerprint impression.

(C) In the Y-axis positive direction extraction, six vectors must be obtained with a predetermined length.

(D) In the Y-axis positive direction extraction, the variance in the direction of the six vectors must be below a certain value.

Coordinate axis quality “OK” This is a case where the coordinate axis is obtained although all the conditions of coordinate axis quality “Excellent” are not satisfied.

Coordinate axis quality “impossible” This is the case where neither the center point nor the Y-axis positive direction is obtained, or only one is obtained.

(3) Fingerprint quality Fingerprint quality is defined by the combination of area quality and coordinate axis quality.

Referring to FIG. 7, the fingerprint quality is determined mainly with the area quality and the coordinate axis quality in consideration of the ease of performing the correction and the data capacity required at the time of the correction. That is, if both the area quality and the coordinate axis quality are “excellent”, it is determined that no correction is necessary, and the fingerprint quality is defined as “excellent”. If the area quality is “excellent” and the coordinate axis quality is “acceptable” or “impossible”, it is determined that only the coordinate axis needs to be corrected, and the fingerprint quality is defined as “good”. On the other hand, if the area quality is "impossible", the fingerprint quality is unconditionally defined as "impossible", and in other cases, the fingerprint quality is defined as "possible".

The fingerprint quality information described above is used to control the flow of fingerprint data. Referring again to FIGS. 1 and 2, the base paper 5 from which the fingerprint data has been read by the fingerprint reading device 2 is delivered to the operator. The operator inputs the reading number printed on the back surface of the received base paper 5 into the fingerprint monitor device 3. The image retrieving unit 1d of the central processing unit 1 accesses the magnetic disk unit 4 and sends fingerprint data matching the input read number to the fingerprint monitor unit 3. The operator confirms and corrects the extracted content regarding the coordinate axis, ZONE, and core line of the fingerprint data displayed on the fingerprint monitor device 3. When the confirmation / correction work is completed, the base paper 5 is temporarily stored and passed to the verification processing of the verification result.

Next, FIG. 8 (a) shows the flow of fingerprint data according to the fingerprint quality.
Also, the description will be made with reference to (b).

The feature point list (including fingerprint quality information) of the fingerprint data of the fingerprint base paper 5 read by the fingerprint reading device 2 is stored in the central processing unit 1.
Is entered in. From the input feature point list, the fingerprint quality is shown as a key and the quality table shown in FIG. The fingerprint data input to the central processing unit 1 is stored in the original image file 4a as it is. The fingerprint data stored in the original image file 4a will later be transferred from the fingerprint monitor device 3 to the central processing unit 1
Will be accessed from.

Corresponding to the fingerprint quality, monitor display information is displayed on the fingerprint monitor device as shown in FIG. If the fingerprint quality is "excellent", the monitor display information is displayed as "unnecessary", and in this case it is not necessary to correct it. If the fingerprint quality is “good”, only the coordinate axis needs to be corrected, and the feature point list (coordinate axis, ZONE) and the core line are displayed as monitor display information. If the fingerprint quality is "OK", it is necessary to correct the coordinate axis, ZONE, and core line, and the feature point list (coordinate axis, Z
ONE), core line and original image are displayed. If the fingerprint quality is "impossible", it cannot be corrected and unnecessary is displayed as monitor display information.

Referring to FIG. 10, the fingerprint reading device 2 stores only the feature point list in the feature point file 8 via the original image file 4a and the core line file 6 for the fingerprint whose fingerprint quality is determined to be “excellent”. It For the fingerprint whose fingerprint quality is determined to be “good”, the feature point list and the core line are displayed on the fingerprint monitor device 3 via the original image file 4a. If the fingerprint data is not corrected in the fingerprint monitor device 3, only the feature point list is stored in the feature point file 8 via the core line file 6. On the other hand, when the fingerprint data is corrected, the feature point list and the core line are sent to the re-feature extraction device 7 via the core line file 6. Then, only the feature point list is output from the re-feature extraction device 7, and this feature point list is stored in the feature point file 8.

For the fingerprint whose fingerprint quality is determined to be “OK”, the feature point list, the core line, and the original image are displayed on the fingerprint monitor device 3 via the original image file 4a. After being displayed on the fingerprint monitor device 3, the original image is discarded, and the flow of fingerprint data is the same as when the fingerprint quality is "good" depending on the presence or absence of correction.

If the fingerprint quality is "impossible", the feature point list is stored in the original image file 4a, and then the fingerprint data disappears.

When the data flow control based on the fingerprint quality is performed as described above, the capacity of the original image file is 70 to 80% compared to the conventional one.
The operator's corrective operation time by the fingerprint monitor is reduced by 60 to 70%.

〔The invention's effect〕

As described above, in the present invention, based on the fingerprint quality,
Since fingerprint data flow control is performed, the amount of fingerprint data can be reduced, the time required for data transfer can be shortened, and the disk capacity for temporary storage can be reduced.

Further, since the fingerprint data that does not need to be corrected is not displayed on the fingerprint monitor device, the operator does not need to judge whether or not the correction is necessary, and the operator does not make an erroneous correction. Therefore, the confirmation and correction time of the fingerprint data is shortened, and the quality of the data does not become non-uniform due to variations in correction due to individual differences.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG.
Figure shows the flow of fingerprint base paper, Fig. 3 (a) and (b)
Is a diagram showing the features of fingerprints and the matching / non-matching of fingerprints,
FIG. 4 (a) is a diagram showing fingerprint data, FIGS. 4 (b) and 4 (c) are diagrams for explaining a feature point list, and FIGS. 4 (d) and 4 (e) are for explaining core lines. FIG. 4, FIGS. 4 (f) and (g) are diagrams for explaining the original image, FIGS. 5 (a), (b), and (c) are diagrams for explaining the area quality, and FIG. (A), (b), and (c) are diagrams for explaining the coordinate axis quality, FIG. 7 is a diagram for explaining the fingerprint quality, and FIG. 8 (a) is a diagram showing the flow of fingerprint data, FIG. 8 (b) is a diagram showing a quality table, FIG. 9 is a diagram showing the relationship between fingerprint quality and interpretation on operator operation and monitor display information, and FIG. 10 is a diagram showing a flow of fingerprint data according to fingerprint quality. Is. 1 ... Central processing unit (CPU), 2 ... Fingerprint reader, 3 ...
Fingerprint monitor device, 4 ... Magnetic disk device, 5 ... Fingerprint base paper.

Continuation of front page (56) Reference JP-A-53-75824 (JP, A) JP-A-55-138174 (JP, A) JP-A-60-221882 (JP, A) JP-A-60-221883 (JP , A) JP-A-60-221880 (JP, A) JP-A-60-221879 (JP, A)

Claims (1)

[Claims] 1. A fingerprint reading unit for extracting a feature point list including fingerprint quality information, a core line, and an original image from a fingerprint image, and a fingerprint quality for determining whether to input the core line or the original image based on the fingerprint quality information. Determining means, and storing the feature point list in the file means,
An output unit that outputs the core line or the original image to the file unit according to the determination of the fingerprint quality determination unit, an image search unit that searches the file unit, a search content searched by the image search unit, and the search It has a fingerprint monitor means for checking and correcting the contents, and according to the fingerprint quality information, no display is required, or the feature point list and the core line, or the feature point list, the core line and the original image are displayed. Fingerprint data flow control device with characteristic fingerprint quality.