JPS61100872A - Method of discriminating individual - Google Patents

Abstract

PURPOSE:To enable to make surer discrimination using fewer information by extracting specified stationary points included in the face of an individual to be discriminated, collating the data of relative position with data stored beforehand and discriminating the individual. CONSTITUTION:A monochromatic CCD camera 12 and a light emitter 13 are provided toward the face 11 of an individual. The picture from the camera 12 is stored in a 1 frame memory 14, and the memory signal is supplied to a microcomputer 15, and at the same time, and so-called strobe lighting is made by the light emitter 13 by command of the microcomputer 15. Corners of eyes R3, L3, R4, L4, tips R2, L2, of left and right ears, nostrils R6, L6, etc. are extracted by an ordinary processing device from the picture stored in an image memory 16, and the individual is discriminated using ratio of the distance A of R3, L3 and distance B of R4, L4 etc.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention is a method of t4 turn recognition, and particularly relates to a method of identifying individuals based on Iruma's face.

[Conventional technology]

Until now, no suitable method has been proposed for analyzing human faces, extracting their features, and identifying individuals. In addition, in the method of processing the entire face image for discrimination, the amount of data is enormous, and the amount of calculation required for verification is also enormous, making it difficult to process with microcomputers. There wasn't.

[Problem that the invention seeks to solve]

Until now, no suitable method has been proposed for identifying individuals based on human faces. Another problem is that the amount of data and calculations is so huge that it cannot be processed by a microcomputer or the like.

[Means for solving problems]

The present invention extracts a predetermined fixed point R5+L5 r R4T L4 included in the face of an individual to be discriminated, and compares the relative position data with pre-stored data to discriminate the individual. This is a determination method.

[Effect]

According to the above method, by making the fixed point clear,
It is possible to perform more reliable discrimination with less information, and it is possible to discriminate between human individuals at the so-called microcomputer level.

〔Example〕

For example, when focusing on the human eye, the outer and inner corners of the eyes are fixed points whose positions do not change even if the facial expression or the like changes. Therefore, for example, as shown in FIG.
The positions of No. 4 are extracted and their relative positional relationships are stored. For example, when a sample as shown in B is supplied, the positions of the fixed points R3°T, , R4, and L4 match, so it is determined that there is a possibility that the individual is the same as the memorized individual, and As shown in C and D, when the positions of the above-mentioned points do not match, it is determined that there is no possibility that the person is the same person and that the person is a different person.

Furthermore, when the ratio of the eye size and width matches the stored data as shown in E, it is determined that there is a possibility that the person is the same person.

Further, FIG. 2 shows a specific example of a configuration for realizing the above method. In the figure, a black and white COD camera (6) and a light emitting device α1 are provided toward the edge of the face of the individual to be discriminated. The image from this camera α is 1 frame memory α4)
This stored signal is stored in the microcomputer α→
At the same time, the light emitter α [present] emits so-called strobe light in response to a command from the computer α9.

(In the computer aυ, for example, a binary image as shown in FIG. 3, which has been subjected to contour detection processing from the signal stored in the memory α→, is stored in the built-in image memory ae,
A fixed point is extracted from this image and a 41'' separation is performed. That is, for example, as shown in FIG.
, the inner corner of the eye R,, in addition to L4, the tips of the left and right ears R21L2,
The nostrils R6, L6, etc. are extracted using general processing means, and the distance A between the outer corners of the eyes R5 and L3 and the distance B between the outer corners of the eyes R4° and L4 are extracted.
Is it possible to identify individuals using the ratio of conduct.

The above-mentioned light emitter (13) is required to obtain a still image with good contrast.The capacity of one frame memory α is 1.2 to 1.6 Mbits, and the capacity of the image memory αQ is
The capacity of about 0.2 Mbit is good.

Furthermore, in addition to the above-mentioned fixed points, each point shown in FIG. 5 is taken, and the position of each point is normalized based on the distance between A and B mentioned above and compared with the stored data.

In this case, matching is performed in descending order of immobility,
Throw away anything that doesn't meet your criteria. Regarding the lips,
Verification is performed by setting a function t (R7L7.8-9).

For example, when using B as the standard, A / B − + a / B − + c / B − + d
/ B −+ b / B→r((R7−L7)/B,
Verification proceeds in the order of (8-9)/B). Furthermore, when comparing, we make judgments based on reality by being stricter on points with high immobility and less lenient on points with low immobility.

Note that instead of the above-mentioned normalization, the zoom of the camera (2) may be controlled by a microcomputer, and feedback may be applied so that A or B becomes the reference length.

〔Effect of the invention〕

According to the present invention, by clarifying the fixed point, it has become possible to perform individual discrimination of human beings at the so-called microcomputer level so that more reliable discrimination can be made with less information.

In addition, since judgments are made based on fixed points, there will be no misjudgment even if the animal is in the process of growth or if its facial expression is intentionally changed.

Furthermore, even if the image is not viewed directly from the front, it can be verified by setting an appropriate proportionality coefficient.

[Brief explanation of drawings]

FIG. 1 is a diagram for explaining the present invention, FIG. 2 is a configuration diagram of an example of an apparatus for realizing the invention, and FIGS. 3 to 5 are diagrams for explaining the invention. R3* L 3 y Ra p, La is a fixed point, (2
) is a camera, α4) is a 1-frame memory, and (2) is a microcomputer.

Claims (1)

[Claims] An individual discrimination method in which a predetermined fixed point included in the face of an individual to be discriminated is extracted, and the relative position data thereof is compared with pre-stored data to discriminate the individual.