JP3505713B2 - Curve identification system - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a locking device that replaces a key or a card in a criminal investigation, a ticket inspection that replaces a ticket or a commuter pass,
Cashless credit transactions in place of credit cards, code numbers and seals, electronic commerce in place of asset management operations such as stock trading that confirmed the person by phone, medical records or social welfare or service industry or administration The present invention relates to a curve identification system that can be applied to a fingerprint collation system as a personal identification means that can be applied to data such as a family register search, identification of a badge and face by visual identification.

[0002]

2. Description of the Related Art In a curve identification system that has been applied to a conventional fingerprint collation system or the like, the image of the entire fingerprint is distinguished by whether the features are likely to appear or not, and the information value is weighted according to the image content. Without doing so, the database and the collation system were configured so as to uniformly collate the entire fingerprint image as if by the naked eye.

[0003]

However, a curved line such as a fingerprint has an image of a specific item having a local singular point and a ridge shape that are not easily disturbed by rotation, movement and noise, that is, an image of a portion having high information value. There are partial images, and it is necessary to make effective use of them during matching, and there is much waste in the method of simply handling them uniformly. Then, in the data storage unit for forming the database that is the basis of the collation system, if even the image data of the portion having a small information value is uniformly stored, the memory capacity is wasted and the cost of the collation system increases and It was a factor of limiting the matching speed.

Further, when the collation work, which is not focused on only the essential points and the unprocessed raw data is not advanced, is unnecessarily advanced during the collation, the collation work is inefficient and a high collation rate is obtained. It was difficult to get. Furthermore, in the case of fingerprint matching, if the ridge-shaped image data is used for the matching work without directly relating to the position of the singularity and the singularity, the relationship with the singularity that should originally contribute to the matching work. Is not used, and therefore, the collation deterioration due to rotation, movement and noise was remarkable.

The present invention eliminates the drawbacks of the conventional curve identification system used for fingerprint collation and the like, does not waste the memory capacity, is efficient only by focusing on the essential points when collating, and the ridge shape. For the image data of, by effectively utilizing the characteristic data consisting only of numerical information indicating the positional relationship with the singular point, avoiding the deterioration of the matching accuracy due to rotation, movement and noise, and proceeding the matching process quickly and accurately. The aim is to provide a curve identification system that can be applied to a fingerprint matching system in a simple and inexpensive manner , with quick results obtained at the highest matching rate in the world .

[0006]

In order to achieve the above object, the invention according to claim 1 has a unit length in the extension direction of a curve (40) starting from a singular point (30) as a data format representing the curve. Trace the strings (41), (42), (43), (44) of each of the measurement points (31) (32) (3)
3) Proceed to the end of the curve (40) while adding (34), and chords (51) (52) (53) from the singular point (30) and each measurement point to the measurement point two units ahead. The measuring means for sequentially measuring the length and the numerical value group of the measured data are used to store, reproduce, or collate the curved shape.

As a result, the three strings (51) (52) are provided if the curve does not reverse its direction of curvature.
From only the length of (53) and the numerical information of the unit length, 1
One singular point (30) and four measurement points (31) (32)
The positional relationship of (33) and (34) can be restored. Then, when this operation is repeated considering the overlapping measurement points,
By connecting the measurement points on the original curve, the original curve can be completely restored or specified if the curve does not reverse its direction of curvature. Moreover, how to express such a curved shape is
Even if you move or rotate the same curve, three strings (51)
(52) Length of (53), one singular point (30) and 4
The positional relationship between the three measurement points (31) (32) (33) (34) is unchanged.

With this method, a curve can be expressed with a small memory capacity. Therefore, if the curve does not reverse its direction of curvature, the length and shape of the curve can be accurately stored, reproduced, or collated only by the numerical value group of the measurement data. Moreover, even if the sample is moved or rotated, the curve can be identified without any trouble. That is, the data format representing this curve is a method of representing a curve shape that is strong against movement and rotation.

In general, a machine is far more advantageous than a human when comparing only a group of numerical values. These things replace the "thing shape pattern recognition," which was thought to be relatively difficult for humans to do with machines, almost completely with machines, making it possible to accurately and quickly create an enormous amount of database. Has a manageable effect.

According to the second aspect of the invention, an image composed of a plurality of pixels that are macroscopically dense and microscopically arranged at regular intervals is converted into the numerical value group. In the image processing algorithm, when the measurement point exists at the position of the interval, a certain reference point (10
0) from the reference point (100) as a means for calculating the coordinates of the measurement point (12) on the curve at a specified linear distance (99). Assuming a triangle connecting a second pixel position (11) farther than a prescribed straight line distance (99) from the pixel position (10) and the reference point (100) and the reference point (100), the reference point (100) and a first side connecting the first pixel position (10), a length of a second side connecting the reference point (100) and the second pixel position (11), and the specified linear distance (99). ) With a measurement point calculation algorithm for performing approximate calculation to specify the coordinates by proportionally distributing the third side connecting the first pixel position (10) and the second pixel position (11). .

As a result, it is possible to calculate a point which is overlooked by being present in a gap between a plurality of pixels having a visual function that are arranged in a gap when viewed in an enlarged scale, that is, an image pickup surface which can have only a finite resolution. Since the above can be precisely estimated, theoretically infinite resolution can be pursued.

According to a third aspect of the invention, a block subdivision means for subdividing the block image into a plurality of divided block images,
First binarization processing means for enhancing the contrast of the block image to binarize it into black and white to obtain a binarized image, and detecting the direction of ridges or ridges of a fingerprint from the binarized image Ridge direction detecting means and the image information of the line along the ridge direction is determined to be valid, the image information not along the ridge direction is determined to be noise, and the noise is removed to improve the image. To obtain a grayscale image improving means, a second binarizing processing means for enhancing the contrast of the improved image to binarize it into black and white to obtain a binarized image, and the ridge line having a plurality of pixel widths. A thinning means for compressing the width toward the center of the pixel width to convert it to a ridge thinned to a single pixel width, and a proximity end point (20) having a ridge in the same direction or a branch point. End point (21) or the end point (2 ) Is regarded as a pseudo singular point generated due to the lack of the ridge due to noise, and is connected, and an end point and a branch point of the ridge after the pseudo singular point correction are true. It was equipped with a singular point extraction means for extracting it as a singular point and applied it to fingerprint matching. By doing so, fingerprint collation can be performed at a high collation rate even for a fingerprint input image including noise.

In the invention according to claim 4 , there are very few singular points such as ridge end points and bifurcation points among individual fingerprints, and fingerprint matching means using only true singular points is used. A means for using a ridge shape having a newly extracted singular point as a starting point, based on the true singular point, as a characteristic point of the fingerprint, where the applicable range is limited. Therefore, points on other ridges associated with a certain feature point (91) are designated as secondary feature points (93) (94) (95), and the secondary feature point (9
3) Ridge shape starting from (94) and (95) (RS
2) Singular point extraction means for adding (RS3) and (RS4) as singular point ridge shapes to the characteristic points of the fingerprint was provided and applied to fingerprint matching. Therefore, by setting a secondary singular point that can identify the starting point from the true singular point and adding the ridge shape starting from that point to the fingerprint information, it is possible to greatly expand the applicable range. I have to. Specifically, there are about 10 fingerprints that could not be registered only with the singularity ridge shape.
Registered 0% and can now collate.

According to a fifth aspect of the present invention, the thinning means and the pseudo singularity correction means form a column connection processing section, and the original black and white image is passed through the column connection processing section to perform the first classification. End point extraction processing for extracting only the end points, the positive / negative inverting means for inverting the signal extracted by the first end point extraction processing in black and white, and the black and white inverted signal for the second time in the cascade connection processing. And a singular point extracting means that can obtain both an end point and a branch point in the ridge of the original black-and-white image by performing a second end point extraction process of extracting the end point of the second classification obtained by passing through the section. None, applied to fingerprint matching.

By doing so, the ridge line that bifurcates is simplified to the utmost by the black-and-white reversal process, and the image constitution is made up of a set of simple curves that neither branch nor intersect. Since fingerprint collation is performed with a simplified image structure, it is optimal for machine collation of a huge number of samples, and high-speed and accurate fingerprint collation is possible with simple equipment including a smaller amount of memory than before.

[0016] The invention according to claim 6, in an image capturing the fingerprint, the noise generated by various factors, ridge or valley lines is divided, it is often a pseudo feature point is generated, the pseudo It is an object to provide an efficient means for distinguishing a feature point from a true feature point.

Therefore, as a means for discriminating between the pseudo feature point and the true feature point of the fingerprint, the branch point (101) of the ridge or valley line.
For each of the three ridges or valleys connected to, the points (102) (103) on the ridges or valleys that are equidistant from the branching point (101) are determined, and the branching point (10
Three points (102) (10) when 1) is regarded as the origin
3) Two-dimensional coordinates of (104) (Xa, Ya), (Xb, Yb), (Xc,
Yc) is calculated, and these three points (102) (103) (10
The product of the horizontal components of the coordinates (Xa, Ya), (Xb, Yb) of the two points at any two points (102) and (103) of 4).
(Xa ・ Xb + product of vertical component (Ya ・ Yb) + (Xa ・ Xb +
If Ya.Yb) is larger than a certain threshold, it is regarded as a true branch point, and if there are no two points with a value larger than the threshold, it is regarded as a pseudo branch point. Further, it is provided with a pseudo branch point discriminating means that can realize the above-mentioned identification purpose with a small calculation amount, and is applied to fingerprint collation. By doing so, it becomes possible to distinguish a pseudo feature point generated by noise from a true singular point with a small amount of calculation and high accuracy.

The invention according to claim 7 is a television camera or a similar fingerprint image input means for photographing a fingerprint image input by the raster scan method, and the fingerprint image input by the fingerprint image input means. Divided into multiple
Block image locally and fragmentally in one frame
Immediately after taking read in, comprising a signal processing unit for point by point processing until the numerical value group, the means for performing numerical processing of the entire fingerprint image by repeating the processing of different blocks images fingerprint images of a plurality of frames, the fingerprint Applied to matching.

By doing so, the raw image data of the entire fingerprint image, which wasted the memory capacity, that is, the image information before the information amount reduction processing is not accumulated,
The system can be configured with a small memory capacity.

In short, since the information amount of one fingerprint image is too large and the memory capacity is wasted, the process of simplifying and digitizing the image configuration to the ultimate according to the purpose of fingerprint collation is performed.
If it is executed step by step for each block portion of the fingerprint block image, a memory capacity for storing useless information is not necessary.

Then, even if the raster scan progresses for the time required for each processing, it is sufficient to skip the fingerprint image during the processing. Then, the block part of the block image required for fingerprint collation is read by waiting for the opportunity of raster scan in the next frame,
The whole image of the fingerprint is obtained.

In this system, fingerprint collation is possible even if the entire image of the fingerprint is not necessarily obtained. Also, that is why this method of reading the block image locally and fragmentally is effective.

Further, the invention according to claim 8 is applied to fingerprint collation by providing an algorithm for reading one fingerprint image within 6/25 seconds required for 6 to 7 frames and completing the numerical processing. . By doing this,
We were able to simplify equipment while maintaining practicability.

This is because, for example, 6 out of the television shooting for continuously shooting 25 or 30 frames per second.
Assuming that the entire image of the fingerprint is obtained in the frame, one fingerprint image is read in 6/25 seconds or 6/30 seconds and the digitization process is completed. This means that the entire system can be simplified and the cost can be reduced as compared with the case where one fingerprint image is held and processed by the photographed data of a digital camera or the like having an unexpectedly large amount of information.

Further, in the invention according to claim 9 , fingerprint ridge pattern, that is, ridge shape of the fingerprint connected to the singular point is defined as fingerprint feature data by the numerical group, and is registered in advance for collation with the fingerprint feature data. Since it is equipped with a collating means for collating with a collated sample or a constructed database, and applied to fingerprint collation, the collation speed and the collation rate can be increased to the world's highest with a small memory capacity, and it is significantly higher than the conventional one. High-speed and accurate matching is possible.

The invention according to claim 10 is divided into a plurality of parts.
Subdivided into block image, improved against noise or
Correction processing, enhancing the contrast of the block image to make it black and white
A computer capable of executing a series of data processing of binarization, collection of fingerprint feature data and collation of 100 million instructions per second operates, and the minimum required fingerprint feature data of 40 to 60 bytes is assigned to each fingerprint image. Further, since it is provided with the collating means and applied to fingerprint collation, the memory area can be significantly reduced and the collation at high speed can be put to practical use with the highest performance in the world .

The invention according to claim 11 is the extraction means for extracting the fingerprint feature data of a minimum required 40 to 60 bytes from the fingerprint image input from the terminal device, and the fingerprint feature data to a communication network. Connect and fingerprint curve
Since it is provided with means for notifying the terminal device of the result of collation in the database which is the basis of the system for identifying and collating, and since it is applied to fingerprint collation, it can be used in a remote place.

Further, the invention according to claim 12 is to search for personal history data of medical or social welfare or service industry or administration or financial institution, lock, various ticket issuing, ticket issuing, ticket inspection, ticket inspection, check, billing, electronic commerce, asset. Since it is arranged in combination with a terminal device or a single device having any of the functions of management operation, and fingerprint collation is used as a personal identification means, conventionally, a card, key, ticket, commuter pass, signature ,seal,
Identity verification, which was determined by the badge, face, and voice of the person, can be completed by only instant fingerprint verification.

Further, the invention according to claim 13 is arranged by a structure to be embedded in the object of secret protection itself as a part of the function of the software secret protection or the design data secret protection function of the manufacturing consignment LSI, and the fingerprint verification is performed. Since this is used as a means for verifying the identity, the cost for secret protection management is reduced.

According to the fourteenth aspect of the present invention, a rule for personal identification is set by logical operation from fingerprint characteristic data of a plurality of fingers. By doing so, it is possible to freely set a plurality of persons as the allowable range of the right holder, such as family limitation.
For example, it can be used for a vehicle driving lock device that complies with special provisions such as family limitation in automobile accident liability insurance. Also, since one person has 20 fingers, the combination calculated by the series has more variety than the recitation number, and the effect of protecting the object to be protected is greatly enhanced.

[0031]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is an explanatory diagram of digitization of characteristic strings showing an embodiment of the present invention. Figure 1
At 40, a ridge 40 formed by a smooth curve from the singular point 30
Is extended. Here, it is assumed that the bending direction is an arc that does not reverse in an S shape. Sampling points (hereinafter, also referred to as "measurement points") 3 are sequentially obtained by tracing the extension direction of the ridge 40 with the chords 41, 42, 43, and 44 having the unit length, one unit length at a time.
Proceed to the end of the ridge 40 while attaching 1, 32, 33, 34. This is a way to measure the distance of curved sections on a map with a divider. Then, based on the definition of the characteristic string, the length of the base 51 of an isosceles triangle in which three consecutive points, that is, 30 and 31 and 32 are connected by a straight line from the singular point 30 to the sampling point 32 two points ahead, is defined as the characteristic string. Extract.

According to the explanation not limited to the isosceles triangle, the singular points 30 to 31, 32, and 33 are sequentially advanced by one unit at a time, and each of the singular points is divided into two points.
Measuring means for measuring the length of each chord up to the unit measurement point 33, 34, 35, that is, the chord length 51 from the singular point 30 to the second measurement point 32, the first measurement point 31 Chord length 52 up to third measuring point 33, second measuring point 32
To the fourth measurement point 34, the length 53 of the chord is sequentially measured. Numerical value group data composed of these measurement results 51, 52, 53 is used as information indicating the curved shape of the ridge 40, and the simple numerical value group data is used as fingerprint characteristic data. The fingerprint characteristic data is compared with the contents of a pre-built database to collate the fingerprint, and the collation speed and the collation rate are increased with a small memory capacity.

Next, FIG. 2 is a diagram in which the curve 40 to be identified is moved and rotated. Here, if the curve 40 is recognized by converting it into the numerical value, the lengths of the strings 51, 52, 53 do not change even if the curve 40 is moved or inverted. Therefore, the shape of the curve 40 can be recognized, stored, reproduced and identified by the numerical data consisting of the lengths of the strings 51, 52, 53, and even if it is upside down as reflected in a mirror,
Matching is possible if the numerical data match.

Next, FIG. 3 is an explanatory view of the principle of the approximate calculation of the coordinate values of the measurement points existing in the gaps between the pixels. On the image pickup surface of an image pickup device such as a digital camera or a video movie, pixels, which are minute sections that respond to light, such as the retina of the human naked eye, are densely arranged in a predetermined number and area. However, even if the pixels are macroscopically dense,
Microscopically, they are regularly and finitely arranged with a predetermined interval.

By the way, in the case of a digital image processed by a computer or the like, as shown in FIG. 3, the data on the screen is represented by the data of only the points on the pixels arranged in a grid pattern. Among these pixels, a pixel with a pattern is a pixel representing a curve. Therefore, the point 12 estimated to be on the curve of, for example, 4 pixels in length from the certain pixel 100 is not always the actual pixel point. Therefore, it is necessary to calculate the measurement point 12 on the curve which is four pixels away from the information on the positions 10 and 11 of the two pixels on the curve which is close to the length of four pixels. . For example, the position of the relevant measurement point can be calculated by linear approximation based on simple proportional distribution. In this way, the pixel position 10,
From the measurement point 12 different from 11, the strings 51, 5 shown in FIG.
The length of 2,53 can be obtained.

Here, means for calculating the coordinates of the measurement point 12 on the curve 40 at a prescribed linear distance 99 from a reference point 100 on the curve 40 when the measurement point 12 exists at the position of the interval. As a measurement point calculation algorithm. Then, even if the imaging surface, which can only have a finite resolution, that is, in the gap portion of a plurality of pixels that have visual functions arranged in a gap and are overlooked when magnified, the points that are overlooked are calculated exactly. If possible, theoretically infinite resolution can be pursued.

Here, a triangle connecting the first pixel position 10 closer to the reference point 100 than the straight line distance 99 and the second pixel position 11 farther from the reference point 100 than the straight line distance 99 to the reference point 100 is. Suppose. And the reference point 10
The line segment connecting 0 and the first pixel position is the first side, the reference point 100
And a second pixel position 11 are referred to as a second side, and a line segment connecting the first pixel position 10 and the second pixel position 11 is referred to as a third side. An approximate calculation is performed to specify the coordinates by proportionally distributing the third side according to the length and the ratio with the linear distance 99. Depending on the accuracy of this approximation calculation, the curve discrimination ability is remarkably improved.

FIG. 4 is an explanatory diagram of a curve shape system using the azimuth angle and the curvature of the characteristic points. At a branch point 401 where the ridge 400 shown in FIG. 4 is branched, consider a ridge 400 extending in the opposite direction of the branch. First, the branch point 401 itself is set as a first locus point, and a point on the ridge 400 that is separated from the first locus point by a distance D is set as a second locus point 402.
Further, a point on the ridge 400 that is also separated from the second locus point 402 by the distance D is set as a third locus point 403. The same operation determines the fourth locus point 404 and the fifth locus point 405. The locus points 401 to 405 here are ridges 400.
The number can be arbitrarily increased until is interrupted.
Next, these locus points 401 to 405 are sequentially divided into straight line segments,
That is, by connecting the strings 501 to 504, the ridge line 400
Straight line segment having the same length D, that is, the chord 501-
It is approximated by a connection of 504.

In order to efficiently express this straight line segment, that is, the curved shape of the ridge 400 approximated by the chords 501 to 504, first, the straight line segment connecting the branch point 401 and the second locus point 402 is the reference coordinate axis 407. Angle (hereinafter referred to as “azimuth angle” of the feature point) is calculated, and then a straight line connecting the branch point 401 and the second locus point 402, and the second locus point 402 and the third locus An angle 409 formed by a straight line connecting the points 403 is calculated (hereinafter, referred to as “first curvature degree” of the feature point),
Further, an angle formed by a straight line connecting the second locus point 402 and the third locus point 403 and a straight line connecting the third locus point 403 and the fourth locus point 404, that is, the second curvature 410, Similarly, the angle formed by two adjacent straight line segments is sequentially obtained. Here, the information necessary to express the curved shape of the ridge is the azimuth 408 of the feature point and a plurality of adjacent straight line segments, that is, the angles formed by the chords 501, 502, 503, and 504. Second and third curvatures 409, 410, 4
Only 11

These characteristic points, that is, the azimuth angle 408 and the curvature degrees 409 to 41 of the branch point 401 and the locus points 402 to 405.
By using the information of 1 as a method of expressing a curved line shape regarding a ridge including a feature point and performing fingerprint matching based on this expression, it can be expected that the identification ability of different fingerprints is significantly improved. It should be noted that the method of expressing the curved shape can be applied to not only the characteristic points where the ridges are branched as shown in FIG. 4 but also the characteristic points including the end points (see FIG. 7) where the ridges are broken.

FIG. 5 shows a true singular point extraction algorithm by thinning and positive / negative inversion. Thinning process (step S
7) and the pseudo singular point correction (step S8) to form a cascade connection processing unit, and the original monochrome image is passed through the cascade connection processing unit. Further, the algorithm for converting a surface into a line is thinned by making the width of one pixel by utilizing only the frame line in the longitudinal direction of the black surface object (enclosed outline outline) in the original black and white image. This is shown as a process (step S7), and the black surface object is simplified to line segments. A specific thinning process (step S7)
The state will be described later with reference to FIG.

Then, the first end point extraction processing for extracting only the end points of the first classification, the positive negative inversion for inverting the signal extracted by the first end point extraction processing in black and white (step S70), and the black and white inversion thereof. It is confirmed by the number of processing times confirmation (step S72) that the received signal has passed through the cascade connection processing section only twice, and a second end point extraction processing for extracting the obtained end points of the second classification is performed. By doing so, it serves as a singular point extracting means that can obtain both an end point and a branch point in the ridge of the original black-and-white image, and can be applied to fingerprint collation. This will be described later in detail with reference to FIG.

By doing so, the ridge line bifurcating is simplified to the extreme by the black-and-white reversal process, and the image constitution is made up of a set of simple curves that neither branch nor intersect. Since fingerprint collation is performed with a simplified image structure, it is optimal for machine collation of a huge number of samples, and high-speed and accurate fingerprint collation is possible with simple equipment including a smaller amount of memory than before.

FIG. 6 shows a true singular point extraction algorithm including a grayscale image improving means. The first binarization process (step S3) is performed to obtain a binarized image by binarizing into black and white by enhancing the contrast of the block image divided into a plurality of subdivisions by the block subdivision means and performing the binarization. Ridge direction detection (step S4) is performed to detect the convex stripe pattern of the fingerprint, that is, the direction of the ridge, from the digitized image.

Then, by detecting the ridge direction (step S4), the direction of the ridge is obtained and used as auxiliary information when removing noise components. That is, the image information of the line along the direction of the ridge is judged to be valid, the image information not along the direction of the ridge is judged to be noise, and the noise is removed to obtain an improved image Grayscale image improvement (Step S5)
Leading to. In addition, binarization is a conventional means for digitizing an analog electric signal having an ambiguous intermediate value, but here,
It is a means for converting a digital electric signal represented by a plurality of bits such as 8 bits into a 1-bit digital electric signal, which is based on a magnitude comparison operation with a certain reference value.

Here, FIG. 7 is an explanatory diagram of the pseudo singular point, and (a) an explanatory diagram of the end points (20) that are close to each other with the ridge in the same direction. (B) It is explanatory drawing of the end point (21) near a branch point. (C) It is explanatory drawing of the end point (22) near a frame of an image. True singular point extraction (step S8) is performed through pseudo singular point correction (step S8) for correcting noise components from the fingerprint data.
9). A specific means of this pseudo-singular point correction (step S8) is post-processing, which will be described with reference to FIG. FIGS. 7A, 7B, and 7C are explanatory diagrams of three types of pseudo singular points that are erroneously captured by noise components and the like and are erroneously recognized as singular points even though they are not singular points. The correction rule that connects the end points close to the frame or the branch point or the end points that are close to each other with a ridge line in the same direction as a pseudo singular point generated due to the lack of the ridge line due to noise is pseudo as the post-processing. It is applied to the singularity correction (step S8).

Returning to FIG. 6 again, the second binarization process (step S6) is performed in which the contrast of the improved image is enhanced to be binarized into black and white to obtain a binarized image.
In the binarized image, the ridge having a plurality of pixel widths, that is, a ridge having a certain width is compressed toward the center of the pixel width, thereby thinning to a single pixel width. A thinning process (step S7) of converting into a ridge is performed. The thinning process (step S7) has the effect of further simplifying the image of a complicated shape and adapting it to the digitization and machine calculation process to simplify the collation work.

Next, the adjacent end points 2 having ridges in the same direction
Pseudo-singular point correction is performed (step S8) in which 0 or the end point 21 close to the branch point, or the end point 22 close to the frame of the image is regarded as a pseudo-singular point generated due to the lack of the ridge due to noise and connected. . As described above, in FIG. 6, the end points and the branch points of the ridge after the pseudo singular point correction is performed by a series of image processing on the input image, and are regarded as true singular points and extracted (step S9). It was equipped with means and was applied to fingerprint verification. By doing this,
Fingerprint matching can be performed at a high matching rate even for a fingerprint input image including noise.

8A and 8B are explanatory diagrams of a singular point extraction method by black-and-white inversion and thinning. FIG. 8A is an explanatory diagram of a black-and-white block image before processing, and FIG. Figure, (c) is an explanatory diagram of a black-and-white inverted block image, (d)
FIG. 4 is an explanatory diagram of an extracted singular point after final processing. Although it is a true singular point extraction algorithm based on the thinning process (step S7) and positive / negative reversal (step S70) already described with reference to FIG.
The image processing from (a) to (b) in (3) is performed, and the amount of information is remarkably reduced as compared with that before the processing, but the digitization of the bifurcation branch remains a burden. In FIG. 8B, the amount of information is considerably reduced as compared with FIG. 8A, but since it has a branch point, it becomes a bottleneck that unique determination cannot be performed at the time of fingerprint collation.

Therefore, in order to eliminate the bifurcation branch point,
Image processing from (a) to (c) in FIG. 8 is performed by positive / negative reversal (step S70), and (c) to (d) in FIG. 8 is further performed by thinning processing (step S7).
Image processing of, the bifurcation point is one end of a finite curve,
That is, it is represented by a singular point. Moreover, this process is thoroughly performed by repeating these processes twice. In this way, if a finite curve without a branch point and a singular point after final processing are extracted, the amount of information can be further reduced by eliminating the branch point, and digitization is easy. The amount is absolutely smaller than that before image processing. Then, fingerprint collation results in a unique judgment item, and it becomes unnecessary to make an artificial or multi-item judgment, so that it becomes suitable for machine processing, the processing speed is increased, and the memory capacity is small.
The burden of constructing and maintaining the database is also reduced.

FIG. 9 is an explanatory diagram of a curve identification system using secondary feature points. As shown in FIG. 9, when focusing on a certain feature point 91, it is on the coordinate axis of the ridge direction axis Y1 connected to that feature point 91 and its vertical axis X1, and at an equal distance from the feature point 91. The four points 92, 93, 94, and 95 are regarded as secondary feature points, and the shapes of ridges RS1, RS2, RS3, and RS4 closest to these secondary feature points are the curved shapes described above with reference to FIG. Extract using the expression method of. These four secondary feature point ridge shapes RS1, RS2, RS
3, RS4 is a fragmentary capture of the pattern of the entire fingerprint viewed from the original feature point 91. By using this additional information for fingerprint matching, the matching accuracy can be significantly improved. Can be expected. Therefore, by setting a secondary singular point that can specify the starting point from the true singular point, and adding the ridge shape having that point as the starting point to the fingerprint information,
This makes it possible to greatly expand the scope of application. Specifically, fingerprints that could not be registered only with the singularity ridge shape can now be registered and compared almost 100%.

FIG. 10 is an explanatory diagram relating to determination of removal of pseudo singular points using inner products. As shown in FIG. 10, the branch point of the true ridge is divided into an area A101 surrounded by one narrow angle and areas A102 and A103 surrounded by two wide angles with the branch point as the center. Most of them have a shape that looks like. Further, the end point of the true ridge can be regarded as a branch point of the valley line, as shown in FIG. 8, and in this case, like the ridge branch point, it is surrounded by one narrow angle. It has the property of having a shape such that it is divided into a region A101 and two regions A102 and A103 surrounded by a wide angle. Using these properties, in order to discriminate the region in which the branch point of such a ridge or valley is divided with a small amount of calculation, as shown in FIG. Point 10 on the ridge or valley
Two-dimensional coordinates (Xa, Ya), (Xb, Yb), (Xc, Yc) of the three points when 2, 103, 104 are obtained and the branch point is regarded as the origin
Ask for.

Then, in the pair of the point 102 and the point 103, the product of the horizontal component (Xa.Xb) and the product of the vertical component (Ya.Y) of the coordinates are obtained.
b) the sum value Request (Xa · Xb + Ya · Yb ) ( hereinafter, the determined Merare Ru value by this manipulation is called "inner product value of the two points").
Since the inner product value of these two points is proportional to the cosine of the angle formed by the straight line connecting the branch points 101 and 102 and the straight line connecting the branch points 101 and 103, if the value is large, these two straight lines Minutes have a narrow angle, and conversely if the value is small, they have a wide angle. Similarly, point 103 and point 1
The inner product value at 04 and the inner product value at the points 104 and 102 are obtained by the same operation. If there is exactly one of these three dot product values that is greater than a certain threshold value, this branch point is divided into one narrow angle area and two wide angle areas described above. It is shown that it is divided, and it is determined that it is a true branch point. In other cases, it is determined that this branch point is a pseudo branch point.

Regarding two fingerprint images, even if these fingerprints are of the same finger, the position and direction of the characteristic point will change depending on the position and angle of the finger. In this case, it is necessary to specify the displacement between the position and the direction, perform the appropriate displacement correction, and then compare the feature points. As described above, in order to specify the displacement between the position and the direction and to perform the appropriate displacement correction, it has been said that a huge amount of calculation has been required in the past. There were big restrictions.

Hereinafter, a method of fingerprint collation that does not require the positional deviation correction with respect to the parallel movement deviation will be described. First, as shown in FIG. 11, a two-dimensional space having a rectangular spread of a certain size is divided into cells, and a writable / readable memory having a two-dimensional memory space in which each cell corresponds to a memory address is created. prepare. Figure 11
Shows a memory space of a two-dimensional array structure consisting of M rows and N columns. At the start of the fingerprint collation process, a predetermined initial value (for example, zero value) is stored in each memory address of this memory.

Next, for a pair of feature points selected one by one from two fingerprints, a similarity index value indicating how similar their shapes are is evaluated. The evaluation function for calculating the similarity index value has a large value (for example, 259) when the similarity is high and a low similarity when the similarity is high, based on the similarity of the ridge direction and the curve shape of the feature point. For, use a function that has the property of outputting a small value (for example, 1).

Then, regarding the above two feature points,
Two-dimensional coordinate values (Xa, Ya) of feature points belonging to one fingerprint
From the two-dimensional coordinate value (Xb,
Find a two-dimensional vector (Xb-Xa, Yb-Ya) connecting up to Yb). This is because the former fingerprint is horizontally (Xb-Xa),
It means that these two feature points are located on the same two-dimensional coordinates when translated by (Yb-Ya) in the vertical direction. If the similarity index value between these two feature points has a high value, the actual parallel displacement of the fingerprint is likely to be the two-dimensional vector (Xb-Xa, Yb-Ya). On the contrary, when the similarity index value is low, the shift of the parallel translation of the actual fingerprint is (Xb-Xa, Yb-Ya).
It is unlikely that

Next, the two-dimensional vector (Xb-Xa, Yb
-When (Ya) is regarded as a two-dimensional coordinate value, this coordinate value is shown in Fig. 1.
Check which cell on the two-dimensional space shown in 1 is included, read the value stored in the memory address corresponding to the cell containing this from the memory, and calculate the similarity between the two feature points. The cumulative value obtained by adding the index values is written again to the same memory address.

FIG. 11 is an explanatory diagram of the storage memory area for the cumulative value of the similarity index values in the collation method that does not require the shift correction related to the parallel movement. In the above description, calculation of a similarity index value of a pair of feature points, calculation of a two-dimensional vector between two feature points, and when the two-dimensional vector is regarded as a two-dimensional coordinate value, the coordinate value is included in the grid of FIG. The value stored in the memory address corresponding to that cell
A series of operations of writing the cumulative value obtained by adding the similarity index value (not the sign) to the same memory address is a combination of some or all of the pairs of feature points selected one from two fingerprints. Repeat about. After that, the maximum one (for example, 259) of the cumulative values of the similarity index values stored in each memory address is set as the comprehensive similarity index value of these two fingerprints. If the comprehensive similarity index value is larger than a certain threshold value, it is determined that the fingerprints are the same. By this method, since the shift correction relating to an arbitrary parallel movement is processed at the same time, it is possible to realize fingerprint matching with extremely small calculation amount.

Up to now, the method of fingerprint collation which does not require the shift correction related to the parallel movement has been described.
A method will be described which enables highly accurate collation even with arbitrary rotation and parallel displacement by performing misregistration correction related to rotation. The deviation correction for a certain rotation angle is
The rotation correction of the angle is applied to all the feature points of one of the two fingerprints to be collated. As described above, the feature point data consists of the two-dimensional coordinate value of the feature point and the feature point shape information, but the two-dimensional coordinate value of the feature point is corrected by rotation conversion with the reference origin as the center. As for the feature point shape information, the same angle is corrected with respect to the feature point azimuth angle. Note that the information expressing the curvature of the feature point shape does not need to be corrected because the rotational movement remains unchanged. In this way, after the rotation correction is applied to all the characteristic points of one of the fingerprints, the collation is performed by the fingerprint collation which does not require the positional deviation correction related to the parallel movement described above.

FIG. 12 shows a singular point extraction algorithm for recognition processing by raster scan and digitization of the entire screen.
Since the television camera image input (step S1) is based on the well-known raster scan television system, a one-frame screen is displayed for a predetermined time, for example, 1/25 second or 30 seconds.
Scanning is performed for a second, each pixel and each block data are recognized in a predetermined order, and partial block data is locally stored without storing all image information of one frame (step S92). After that, the singular point extraction process (step S93) shown in FIGS. 5 and 6 is performed in a limited short time.

However, singular point extraction processing (step S9)
In 3), a computer capable of executing 100 million instructions per second operates, and assuming that the fingerprint feature data of about 50 bytes is assigned to each fingerprint image, one fingerprint image is subjected to singular point extraction processing (step S93) and singularity. It takes about ⅕ second to complete the point data storage (step S94) locally and sequentially. Incidentally, the above-mentioned about 1/5 second is actually 0.24 seconds, but for convenience of explanation, it is displayed as a fraction, and 6/25 frames or 6/30 to 30 frames are scanned. This is the time it takes. Therefore,
A raster scan of 5 or 6 to 7 frames corresponds to about ⅕ second, so that the singular point extraction processing (step S93) and the singular point data storage (step S94) are performed over the entire fingerprint image. ) Is completed, the confirmation of the full screen processing (step S95) is determined to be completed, and the recognition processing of one fingerprint image is completed.

FIG. 13 shows an algorithm of the fingerprint collation system shown as an embodiment of the present invention. The algorithms from the TV camera image input (step S1) to the true singular point extraction (step S9) and the confirmation of the full-screen processing (step S95) have already been outlined with reference to FIGS. 5, 6 and 9. It is a concatenation of the contents.

In addition, the television camera image input (step S
Block segmentation immediately after 1) (step S2) is an essential requirement in this embodiment, and the block image segmented into a plurality of segments is binarized (step S3) and grayscale image improved (step S5). Lead to.
If the result of the confirmation of the full screen processing (step S95) is "Yes", the collation processing (step S74) or the storage in the fingerprint collation database (step S75) is reached. In the collation process (step S74), the collated sample data in the fingerprint collation database is collated and collated in the state of digitized data.

It should be noted that since grid noise is generated at the joints of the block images (not shown), the noise is removed by software, but even if this is done, the necessary information will not be lost and all In order to secure or supplement the information for the screen, adjacent blocks are overlapped with each other.

In this way, the computer that executes 100 million instructions per second performs a series of noise removal, image improvement, binarization, thinning, pseudo-singular point removal, and collation with a database of the input fingerprint image. Is performed in 0.24 seconds, and the fingerprint feature data for one fingerprint image is 40
The data amount of 60 to 60 bytes is processed so that it can be identified.

The image quality improvement and various processes will be further described. As shown in FIGS. 6 and 13, the ridge direction is detected from the fingerprint image before the image division (step S4).
The information of the ridge direction is extracted by the above, and the directional filter (not shown) is controlled and used for the gray scale image improvement (step S5). Here, if the fingerprint image is input to the directional filter and there is a portion that needs to be corrected so as to deviate from the general shape of the fingerprint, there is a fingerprint-like shape along the ridge direction information. The information is corrected so that a traced line is added to improve a clearer image. In short, a signal processor (hereinafter also referred to as a DSP) is used to automatically execute a correction, which is equivalent to finding a correction point with the naked eye of a person and speculating from the ridge direction around the correction point to add a hand correction. (Referred to as) program.

In the block subdivision, if a raster scan type television screen is divided into a grid pattern and data processing is performed for each block image, processing of complicated image information, which is not good for a machine, is simplified. , Unique processing can be approached. This is because if one fingerprint image is enlarged, its ridge shape will be simplified, and a line drawing that can be simply drawn only in relation to singular points will be possible, and recognition by machine unique processing will be possible. .

Therefore, if one fingerprint image is enlarged on the television screen and divided into each block image, the unique processing can be approximated. However, not all of the information of each block image is stored, and even local singular points 30 that contribute to fingerprint matching and are resistant to rotation, movement, and noise, and the characteristic strings 51, 52, 53 related thereto are extracted. To
Immediately after reading each block image, the information of the block image corresponding to the image reading unit (not shown) moved while the raster scan progresses while processing the locally completed data each time is performed, and I'm skipping over. In short, I don't read while thinking. By doing so, the required memory capacity can be reduced. Specifically, it is a system configuration in which a small television camera and a signal processor (DSP) not shown are connected. The DSP program is provided with the algorithm of the present invention in the form of a program.

The front screen consisting of 200 × 200 pixels captured in the raster scan system from the small television camera is set to 3
Each small block image of 2 × 22 pixels is temporarily stored in the data memory of the DSP and immediately measured points 31, 32, 33, 34
Are sampled and set along the ridge 40, and are digitized into singular point ridge information, that is, fingerprint characteristic data. In this way, the digitized and extracted fingerprint feature data is stored in a data memory (not shown).

Then, the same processing is performed for the data of the small block image to be input next. Further, since it takes time for the DSP to extract the characteristic strings 51, 52, 53, it is impossible to process consecutive blocks. That is, the computer is busy thinking about the singularity ridge information, and new information is not read during that time to save the memory capacity.

In this system, since the processing of the small block image can be independently executed, for example, 25 frames per second (PAL or SECAM system television in Europe and Russia)
By scanning at a speed of, a fingerprint image of 6 frames is input, and the entire image area is covered as a whole. Conversely speaking, image data of a little more than 1/6 (overlap) of the entire image area is read from the fingerprint image of one frame.

Even if an attempt is made to cover the entire image area from a fingerprint image of one frame, the DSP can digitize the minimum singular point ridge information necessary for fingerprint matching by 0.24.
Since it takes about a second, scan one frame 1/2
It cannot be covered in 5 seconds or 1/30 seconds (NTSC system TV in the US and Japan). Therefore, a set of fingerprint image information is intermittently captured for each of the small block images over a time period of 6 to 7 frames, that is, 6/25 seconds or 7/30 seconds,
Each time it is processed, it is made into a database as substantially unique numerical information, that is, fingerprint characteristic data, which is convenient for the machine to perform fingerprint matching.

Here, in the actual operation method of the fingerprint collation system, first, a large number of fingerprint characteristic data to be collated is registered in the database of the host computer or the information center. Next, fingerprint feature data is extracted from the fingerprint image taken for identification and used for fingerprint matching. A large number of fingerprint feature data registered in the database are compared with each other, and “true” or “false” is determined according to a predetermined comparison degree determination standard.

For example, as a measure of the correctness of fingerprint collation,
As the probability of correct matching, there is a “matching rate”, which is “true”.
It is the percentage of "true" fingerprints contained in the sample judged to be "ideal", and ideally 100%, but if we stick to 100%, perfect fingerprint matching without any noise component Otherwise, it will not be possible to judge, and there is a high probability that the registered fingerprint cannot be accepted.

Further, in addition to that, there is a “mismatch rate” for erroneously determining “false” even if the fingerprint is “true”, and it is included in the samples determined to be “false”. , The percentage of “true” fingerprints, and ideally 0%, but if the “verification rate” is set to 100%, the “non-collation rate” of about several percent is obtained. These concepts are similar to the "true / false judgment matching sensitivity" setting found in the "true" or "false" judgment matching system of bills in vending machines, etc. Is made.

According to the present invention, the "collation rate" is 100%.
Even if it was set to, the "mismatch rate" was successfully suppressed to only 1.2%. This is an epoch-making success compared with the same "mismatch rate" of the world's best fingerprint matching system of 27.72%. This is especially true for rotation,
It also means that the identification ability of the input fingerprint image with movement and noise is improved.

Further, a computer having a signal processing processor capable of executing 100 million instructions per second operates a series of data processing for collecting and collating the fingerprint characteristic data, and the fingerprint characteristic data of about 50 bytes per fingerprint image is operated. If it is identified by the matching means assigned with, the series of operations of noise removal, image improvement, binarization, subdivision, and matching of the input image can be processed in 0.24 seconds. This means that, as a practical level, collation was significantly faster and more reliable than conventional ones.

Also, by connecting to a communication network and communicating information with an information center equipped with a database, it is possible to confirm the identity at a remote location.

Further, the fingerprint characteristic data of a plurality of fingers is set by the logical operation of the fingerprint characteristic data of a plurality of fingers in a combination of AND, NAND, OR, NOR and the like. In this way, the setting that allows the fingerprint data of each thumb of all fathers, mothers, and children to be used by OR is suitable for locking common to families, and it is suitable for multiple people such as family locks as well as home locks. Can be freely set as the permissible range of the right holder, and can be used, for example, in a vehicle driving lock device that complies with special provisions such as family limitation in automobile accident liability insurance. Also, since one person has 20 fingers, the combination calculated by the series has more variety than the recitation number, and the effect of protecting the object to be protected is greatly enhanced.
For example, if the AND of the right thumb and the left index finger is matched, the condition of permission is set. If a confidentiality officer intentionally or negligently leaks fingerprint data, or if the fingerprint data is stolen, if the person perceives that fact, immediately change the combination to AND of the left and right little fingers, etc. Can prevent expansion.

Under the amended law in electronic commerce in the United States, even if the environment in which "a PIN code has the same legal effect as a signature" is reached, the common sense and ultimate judgment regarding the personal identification function are as follows: It is inevitable that the fingerprints will be ranked higher than the code numbers.

[0082]

Since the curve identification system is configured as described above, the invention according to claim 1 can represent a curve with a small memory capacity. Therefore, if the curve does not reverse its direction of curvature, the length and shape of the curve can be accurately stored, reproduced, or collated only by the numerical value group of the measurement data. Moreover, even if the sample is moved or rotated, the curve can be identified without any trouble. That is, the data format representing this curve is a method of representing a curve shape that is strong against movement and rotation. Generally, machines are far more advantageous than humans when comparing only numerical groups. It was said that these things were relatively difficult for humans to do for machines, but "recognizing the shape pattern of things"
Is almost completely replaced by a machine, and it has the effect of processing a huge amount of database, which is still redundant, at high speed and accurately.

According to the second aspect of the present invention, there is an image pickup surface or the like which can have only a finite resolution, that is, it exists in a gap portion of a plurality of pixels having a visual function arranged in a gap when enlarged. Since the points that are overlooked can be calculated exactly, the theoretically infinite resolution can be pursued.

Further, according to the third aspect of the present invention, fingerprint collation can be performed at a high collation rate even for a fingerprint input image including noise.

[0085] Further, according to the invention according to claim 4, specifically TenTakashisen shape alone almost 100 percent fingerprints could not be registered
You can now register and collate.

Further, according to the invention of claim 5 , it is most suitable for machine collation of an enormous number of samples, and high-speed and accurate fingerprint collation can be performed with simple equipment including a smaller memory amount than the conventional one. Become.

According to the invention of claim 6 , it is possible to distinguish a pseudo feature point caused by noise from a true singular point with a small amount of calculation and high accuracy.

According to the seventh aspect of the present invention, the raw memory image data of the entire fingerprint image, which wasted the memory capacity, that is, the image information before the information amount reduction processing is not accumulated, and the small memory capacity is used. The system can be configured with.

Further, according to the invention of claim 8 , simplification of equipment can be realized while maintaining practicality.

Further, according to the invention of claim 9 , the collation speed and the collation rate can be increased to the highest in the world with a small memory capacity, which is significantly faster and more accurate than the conventional one. You can collate.

Further, according to the invention of claim 10 , the memory area is further reduced significantly and the collation at high speed is the highest in the world.
It can be put to practical use in terms of performance .

Further, according to the invention of claim 11 , fingerprint collation at a remote place can be used.

According to the twelfth aspect of the present invention, only the instantaneous fingerprint collation can be used for personal identification, which was conventionally discriminated by a card, a key, a ticket, a commuter pass, a signature, a seal, a badge, a face and the voice of the person. Can be done.

According to the thirteenth aspect of the invention, the cost of secret protection management for the object of secret protection can be reduced.

According to the fourteenth aspect of the present invention, it is possible to freely set a plurality of persons as the allowable range of the right holder, such as family limitation. For example, it can be used for a vehicle driving lock device that complies with special provisions such as family limitation in automobile accident liability insurance. Also, since one person has 20 fingers, the combination calculated by the series has more variety than the recitation number, and the effect of protecting the object to be protected is greatly enhanced.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of digitization of characteristic strings.

FIG. 2 is a diagram in which a curve to be identified is moved and rotated.

FIG. 3 is an explanatory diagram of a principle of approximately calculating coordinate values of measurement points existing in a gap between pixels.

FIG. 4 is an explanatory diagram of a curve shape system using azimuth and curvature of feature points.

FIG. 5 is a true singular point extraction algorithm by thinning and positive / negative inversion.

FIG. 6 is a true singular point extraction algorithm including a grayscale image improving means.

FIG. 7 is an explanatory diagram of pseudo singular points, and (a) is an explanatory diagram of end points that are close to each other with ridges in the same direction. (B) It is explanatory drawing of the end point near a branch point. (C) It is explanatory drawing of the end point near the frame of an image.

FIG. 8 is an explanatory diagram of a singular point extraction method by black-and-white inversion and thinning, and (a) is an explanatory diagram of a black-and-white block image before processing. (B) It is explanatory drawing of the extraction singular point of a halfway process. (C) It is explanatory drawing of a black-and-white inversion block image. (D) It is explanatory drawing of the extraction singular point after the final process.

FIG. 9 is an explanatory diagram of a curve identification system using a secondary feature point.

FIG. 10 is an explanatory diagram of determination of removal of a pseudo singular point using an inner product.

FIG. 11 is an explanatory diagram of a storage memory area for a cumulative value of similarity index values in a matching method that does not require a shift correction related to parallel movement.

FIG. 12 is a singular point extraction algorithm of recognition processing by raster scan and digitization of the entire screen.

FIG. 13 is an algorithm of the fingerprint matching system.

[Explanation of symbols]

10 First pixel position 11 Second pixel position 12 measurement points 20, 21, 22 endpoints 30 singular points 40,400 Curve or ridge 41, 42, 43, 44 unit length strings 12, 31, 32, 33, 34 Measuring points 51, 52, 53 characteristic strings 91 Feature points 92-95 secondary feature points 99 prescribed straight line distance 100 reference points 101,401 branch point 102, 103, 104 Points on ridges or valleys 402-405 locus points 407 Reference axis 408 Azimuth of feature point 409,410,411 Curvature 501-504 length D strings RS1 to RS4 Shape of ridge starting from secondary feature point

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Tomohiko Otsuka 1220-2 Kasumita-cho, Hachioji-shi, Tokyo Tokyo National College of Technology, Department of Electrical Engineering (72) Inventor Mohamed Mustafa 2-chome, Ookayama, Meguro-ku, Tokyo No. 12-1 Tokyo Institute of Technology Graduate School of Science and Engineering, Department of Electrical and Electronic Engineering (56) Reference Ryuzo Takiyama, 2 persons, Measurement and processing of Jomon pottery image by contour information, IEICE technical report PRU86-59-64, November 28, 1986, Vol. 86, No. 251, pp. 25-32 Naoki Ono, 1st person, Equilateral polygon approximation of line figure and its application to recognition, IEICE Technical Report PRU89-89-100, January 25, 1990, Vol. 89, No. 389, pp. 25-31 Kazuhiro Otani, 3 persons, Feature extraction of fingerprint images by the linear sweep search method, Proc. Of the 2000 IEICE General Conference Information and Systems 2, March 7, 2000, p. 210 (58) Fields surveyed (Int.Cl. ⁷ , DB name) G06T ⁷ /00-7/60 G06T 1/00 G06T 3/00 H04N 1/41-1/419 JSTPlus file (JOIS)

Claims (14)

(57) [Claims] 1. A singular point (3
The extension direction of the curve (40) starting from 0) is traced by the chords (41) (42) (43) (44) of the unit length, and the measuring points (31) (32) (33) ( 34) While advancing to the end of the curve (40), the length of the chords (51) (52) (53) from the singularity (30) and each measurement point to the measurement point two units ahead A curve identification system characterized by sequentially measuring, and storing, reproducing, or collating a curve shape by a numerical value group of the measurement data. 2. An image processing algorithm for converting an image consisting of a plurality of pixels which are macroscopically dense and microscopically regularly arranged at a predetermined interval into the numerical value group, wherein measurement points ( As a means for calculating the coordinates of the measurement point (12) on the curve at a specified straight line distance (99) from a certain reference point (100) on the curve when 12) exists at the position of the interval, A first pixel position (1 that is closer than a specified straight line distance (99) from the reference point (100).
0) and the specified straight line distance (9) from the reference point (100).
9) Assuming a triangle connecting the second pixel position (11) farther than the reference point (100) and the reference point (1)
00) and the first side connecting the first pixel position (10), the length of the second side connecting the reference point (100) and the second pixel position (11), and the specified straight line distance (99). The first pixel position (10) and the second pixel position (11)
The curve identification system according to claim 1, further comprising a measurement point calculation algorithm for performing an approximate calculation to identify the coordinates of the measurement point (12) by proportionally distributing the third side connecting the points. 3. A block subdivision means for subdividing into a plurality of divided block images, and binarization processing image by binarizing into black and white by enhancing the contrast of the block image.
The binarization processing means, the ridge direction detection means for detecting the ridge direction of the fingerprint , that is, the direction of the ridge from the binarized image, and the image information of the line along the direction of the ridge. Judgment, judging that the image information that does not follow the direction of the ridge is noise,
Grayscale image improving means for removing the noise to obtain an improved image; second binarizing processing means for binarizing the improved image to black and white to obtain a binarized image; and a plurality of pixels Thinning means for converting the ridge line having a width into a ridge line thinned to a single pixel width by compressing the width toward the center of the pixel width, and an adjacent end point having a ridge line in the same direction. (20) or the end point (21) close to the branch point, or the end point (22) close to the frame of the image,
Pseudo-singular point correction means that regards and connects as a pseudo-singular point generated due to the lack of the ridge due to noise, and the end points and branch points of the ridge after the pseudo-singular point correction are regarded as true singular points. The curve identification system according to claim 1 or 2 , further comprising a singular point extraction means for considering and extracting, which is applied to fingerprint matching. 4. Points on other ridges associated with a certain feature point (91) are secondary feature points (92) (93) (94) (9).
5) and its secondary feature points (92) (93) (94) (9
5) Ridge shape starting from (RS1) (RS2) (R
The curve identification system according to claim 1 or 2 , further comprising: a singular point extraction unit that adds S3) (RS4) to the characteristic points of the fingerprint as a singular point ridge shape, and is applied to fingerprint matching. 5. The thinning means and the quasi-singular point correction means form a cascade connection processing unit, and the original monochrome image is passed through the cascade connection processing unit to extract only the end points of the first classification. The first end point extraction processing, the positive / negative inverting means for inverting the signal extracted by the first end point extraction processing in black and white, and the black and white inverted signal passed through the cascade connection processing section in the second time and obtained. By applying a second end point extraction process of extracting the end points of the two classifications, a singular point extracting means that can obtain both end points and branch points in the ridge of the original black and white image is applied to fingerprint collation. The curve identification system according to claim 3 , wherein: 6. As a means for discriminating between a pseudo feature point and a true feature point of a fingerprint, for each of three ridges or valleys connected to the ridge or valley branch point (101), the branch point ( The points (102) (103) on the ridge line or the valley line that are equidistant from 101) are obtained, and three points (102) (103) (1) when the branch point (101) is regarded as the origin.
04) two-dimensional coordinates (Xa, Ya), (Xb, Yb), (Xc, Yc) are obtained, and any two points (102) (102) (of these three points (102) (103) (104) ( 103), the product of the horizontal component (Xa ・ Xb) and the vertical component (Ya ・ Yb) of the coordinates (Xa, Ya), (Xb, Yb) of the two points is added (Xa ・ Xb + Ya ・If Yb) is larger than a certain threshold value, it is regarded as a true branch point, and if there are no two points having a value larger than the threshold value, it is regarded as a pseudo branch point. curve identification system according to any one of claims 3 to 5 in computational comprises a pseudo-branch point determination means that allows realizing the identifying objects, characterized by being applied to a fingerprint matching. 7. A television camera or an equivalent fingerprint image input means for photographing a fingerprint image input by a raster scan method, and a fingerprint image input by these fingerprint image input means are divided into a plurality of block images. immediately after local and fractionally take read in one frame, a signal processing unit for point by point processing until the numerical value group, digitizing of the entire fingerprint image by repeating the processing of different blocks images fingerprint images of a plurality of frames The curve identification system according to any one of claims 1 to 5 , wherein the curve identification system is provided with a processing unit and is applied to fingerprint collation. 8. The method of claim 7, characterized in that is applied to the fingerprint collation includes an algorithm completed to the reading and digitizing process one fingerprint image within 6 seconds duration 25 minutes 6-7 frames The curve identification system described in. 9. A ridge pattern of a fingerprint connected to the singular point, that is,
The fingerprint feature data in which the shape of the ridge is specified by the numerical value group is provided, and a matching means for matching with a pre-registered matching sample or a built-in database for matching the fingerprint feature data is provided and applied to fingerprint matching. The curve identification system according to any one of claims 1 to 8 , characterized in that: 10. A sub-image is divided into a plurality of block images.
Processing, improvement or correction processing for noise, the block image
A computer that can execute 100 million instructions per second to perform a series of data processing such as binarization into black and white by enhancing the contrast of fingerprints , collection and verification of fingerprint characteristic data, and a minimum of 40 to 60 bytes per fingerprint image are required. The curve identification system according to any one of claims 1 to 9 , further comprising the collating unit to which the fingerprint feature data is assigned, which is applied to fingerprint collation. 11. An extracting unit for extracting the minimum required 40 to 60 bytes of the fingerprint feature data from the fingerprint image input from the terminal, and connecting the fingerprint feature data to a communication network to identify a fingerprint curve. Of the matching system
Curve identification system according to any one of claims 1 to 10 the results of collation of the underlying database comprises means for notifying the terminal device, characterized by being applied to a fingerprint matching. [Claim 12] Any one of medical history, social welfare, service industry, administration or financial institution's personal history data search, locking, various ticket issuing, ticket issuing, ticket inspection, ticket inspection, checkup, charging, electronic commerce, asset management operation disposed in combination with the terminal device or a single device having the features of claim 3 to claim 1, characterized by using the fingerprint collation as identification means
Curve identification system according to any one of 1. 13. As a part of the function of software confidentiality protection or design data confidentiality protection of a manufacturing consignment LSI, the structure is embedded in the object of confidentiality protection itself, and fingerprint collation is used as a personal identification means. curve identification system according to any one of claims 3 to 12, characterized in. 14. The curve identification system according to any one of claims 3 to 13 and sets the identification rules by logical operation from the fingerprint characteristic data of a plurality of fingers.