JPH03218574A - Feature extraction device - Google Patents

Abstract

PURPOSE:To register only the fingerprint of good quality with sufficient number of features effective for collation processing, and to improve collating precision by investigating whether the number of the features extracted from the registered finger-print is more than a threshold determined beforehand or not. CONSTITUTION:Means 10, 11 to pickup the figerprint picture of a finger placed on a transparent body and output it as two-dimensional quantized data, the means 20 to extract the feature from the quantized data corresponding to all the fingers according as plural fingers are placed successively on the transparent body, the means 15 to decide whether the features effective for collation are extracted by more than the threshold determined beforehand or not, and the means 24 to register feature information about the fingerprint with the feature more than the threshold together with personal identification information are provided. Namely, the feature deciding part 15 of a deciding means decides to register the finger-print when the number of pieces of feature point in a feature point list generated by the feature extracting part 20 is more than enough to the collation processing, and stores the feature point list in the file 24. The fingerprint feature of good quality can be extracted effectively, and the characteristic of a collated result can be improved.

Description

[Detailed Description of the Invention] Technical Field The present invention relates to a feature extraction device, and more particularly, to a feature extraction device that extracts a feature file necessary for a matching process in a fingerprint matching system that determines the identity of a human fingerprint and a registered fingerprint based on the fingerprint characteristics and performs personal identification. The present invention relates to a feature extraction device for creating features.

Conventional technology Fingerprint matching systems that use fingerprints to identify individuals have extremely high reliability due to the characteristics of fingerprints, which are ``unique for all people'' and ``unchangeable throughout life.'' Devices for automatically comparing a collected fingerprint with a pre-registered registered fingerprint have already been proposed, and there are many techniques including Japanese Patent Laid-Open No. 59778.

On the other hand, with the recent development of the information society, the importance of personal identification is increasing when logging on from a computer terminal, accessing a database, and managing entry to seven important confidential areas. Currently, many of these are recited numbers or I+)
This is done using cards, etc., but these easily lose their effectiveness due to theft, loss, forgery, etc., and it must be said that their reliability is low.

Against this background, the realization of a personal identification device using fingerprint matching technology is eagerly awaited. In this type of device, when inputting a fingerprint, it is not desirable for operational reasons to apply ink to the fingerprint, press it onto a printing paper, and input the ink fingerprint image, as in the case of criminal investigation. Japanese Patent Laid-Open No. 54-69300 proposes a method of inputting a fingerprint image using optical conditions by pressing a finger on a glass surface without using ink. However, when pressing a finger on a glass surface, ridges may coalesce due to uneven pressing pressure, and the difference in height of the unevenness of the ridges is small, resulting in low-quality fingerprints with little difference in shading. There is a problem that is collected.

As mentioned above, among the fingerprints input as registered fingerprints,
The fingerprints include low-quality fingerprints, and feature extraction from these low-quality fingerprints is generally difficult, which in turn leads to a deterioration of the verification results.

Purpose of the Invention Therefore, the present invention has been made to overcome the drawbacks of the prior art, and its purpose is to effectively extract high-quality fingerprint features and improve the characteristics of the verification results. An object of the present invention is to provide a feature extraction device that achieves this.

Another object of the present invention is to provide a feature extraction device that can sufficiently handle a large number of fingerprint registrations and improve operational efficiency.

Structure of the Invention According to the present invention, a human fingerprint image and a registered fingerprint image are compared to extract fingerprint pattern features used in checking whether the human fingerprint image matches the registered fingerprint image. A feature extraction device, comprising means for collecting a fingerprint image of a finger placed on a transparent body and outputting it as two-dimensional quantized data, and responding to a plurality of fingers being sequentially placed on the transparent body. means for extracting features from the quantized data corresponding to all the fingers; means for determining whether features effective for the verification have been extracted in excess of a predetermined threshold; and means for extracting features from the quantized data corresponding to all fingers; There is obtained a feature extraction device characterized in that it includes means for registering feature information together with personal identification information.

Example Next, the present invention will be explained with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. This embodiment includes a fingerprint image input unit 10 that collects a gray-scale fingerprint image of a finger placed on a transparent body and photoelectrically converts the image, and an A/D converter 11 that converts the photoelectric conversion signal into two-dimensional quantized image data.
and a fingerprint placement inspection unit 1 that detects placement of a finger and outputs a detection signal by converting the density of pixels at predetermined positions in the collected image.
2, an input image acquisition control unit 13 that performs input image acquisition control after receiving a detection signal, an image storage unit 14 that stores a two-dimensional quantized image of the acquired fingerprint, and an input image acquisition control unit 13 that performs input image acquisition control after receiving a detection signal; A feature determination unit 15 that determines whether the value is above a certain value, a file 24 that stores extracted features, and various information such as an ID code (personal identification information) to be given to each feature file when creating a feature file. An input unit 23 that inputs input, an output unit 22 that outputs the feature extraction process and the feature determination results,
It consists of a feature extraction section 20 that performs feature extraction processing.

Note that 23 is an image output section.

Regarding the fingerprint image input unit 10, a process of inputting a light source converted image of a fingerprint pattern using an imaging device using optical conditions caused by incident light from a light source with respect to a finger placed on a transparent body. For example, JP-A-54-1
There is a device described in Japanese Patent No. 39300 and is used.

The fingerprint placement inspection unit 12 and the input image acquisition control unit 13 detect the placement of a finger by converting the gray scale values of pixels at predetermined positions in images repeatedly captured from the imaging device, and input detection signals into the input image. The image is sent to the collection control unit 13, thereby collecting the image input.
There is a device described in Japanese Patent No. 255480, which is used. The two-dimensional quantized image thus obtained is written in the image storage section 14.

FIG. 2 shows a fingerprint pattern button including minutiae points (hereinafter referred to as maneuvers) in this embodiment, FIG. 3 shows directions of end points and branch points, and FIG. 4 shows a schematic configuration diagram of a manualist for each maneuver.

In Figure 2.3.4, the fingerprint features are the endpoint 2ol,
Position coordinates of minutiae Mi at branch point 202 (Xi, Y
i), direction Di 204, four-neighborhood Mayusha and ration (rl, r2, r3, r4) 203 and?
-1- This is a list representation of a plurality of minutiae Mi that exist in one fingerprint, using the type qi of Yusha as a unit. These detailed definitions and detection methods are described in Japanese Patent Application Laid-open No. 138174/1983, and their explanation will be omitted here.

Basically, the verification process is performed based on whether or not the above-mentioned minutiae is present in the fingerprint pattern, and if a certain number or more of minutiae are confirmed on the pattern, the fingerprints are considered to be the same. This verification method is also described in detail in Japanese Unexamined Patent Publication No. 55-138174, and therefore will not be described here.

The feature determining unit 15 in FIG. 1 determines to register if the number of minutiae in the manualist created by the feature extracting unit 2o or the number of minutiae is sufficient for the matching process, and the above-mentioned manualist is registered. It is stored in the file 24.

FIG. 5 is a diagram showing the processing flow in this apparatus. The finger placed on the fingerprint image input section 1o is detected by the fingerprint placement detection section 1.
In step 2, the input timing is controlled by examining temporal gradation changes on the image, and the images are stored one after another in the image storage section 14. A message from the output unit 22 asks whether all fingers to be registered have been input, and a determination is made as to whether the capacity of the image storage unit 14, for example, the hard disk, is full. If it is full, An input impossible message is displayed on the output section 22.

Regardless of whether the capacity of the image storage unit 14 is full or not, after the registered finger is input, the above-described feature extraction and feature determination are sequentially performed on the fingerprint stored in the image storage unit 14. If registration is possible, it is registered in the file 24. If registration is not possible, registration is not performed and the next fingerprint is processed. After all the fingerprints have been processed, a message to that effect is displayed on the output section 22, and the input section 23 gives each mechanicalist an ID code for identifying the individual who has the fingerprint.

After the manualist is stored in file 24,
In response to a command from the central processing unit 17, a storage completion message is displayed on the output unit 22, and then an ID number is assigned to the stored manualist via the input unit 23.

In this embodiment, the number of minutiae is used as a criterion for determining whether or not to register in the feature determination unit 15, but various features that occur in the feature extraction process, such as low image quality in local regions Even if the number of "unknown areas" indicating the number of unknown areas is used, the essence of this process remains unchanged.

Effects of the Invention As explained above, the present invention determines whether the number of features extracted from registered fingerprints is greater than or equal to a predetermined threshold, thereby determining the number of features that are effective for verification processing. This has the effect of registering only high-quality fingerprints and improving matching accuracy.

In addition, multiple fingerprints are stored on a storage medium such as a disk,
Performing batch feature extraction processing when the system is not in operation has the effect of significantly improving operational efficiency when registering a large number of fingerprints.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 is a schematic diagram of a minutiae in the embodiment of Fig. 1, and Fig. 3 is a schematic diagram of the direction of the end point and branch point in the embodiment of Fig. 1. Figure 4 shows the manualist, Figure 5 shows the first
FIG. 3 is a process flow diagram in the illustrated embodiment. Explanation of symbols of main parts 10...Fingerprint image input unit 11...A/D conversion unit 12...Fingerprint placement inspection unit 13...Input image collection Control unit 14...Image storage unit 15...Feature determination unit 20...Feature extraction unit 24...File

Claims (1)

[Claims] (1) A feature extraction device that compares an input fingerprint image and a registered fingerprint image and extracts fingerprint pattern features used in checking whether the input fingerprint image matches the registered fingerprint image. a means for collecting a fingerprint image of a finger placed on a transparent body and outputting it as two-dimensional quantized data; a means for extracting features from the quantized data corresponding to the quantized data; a means for determining whether features effective for the verification have been extracted above a predetermined threshold; A feature extraction device characterized in that it includes means for registering the same.