JP2551191B2 - Fingerprint pattern matching method - Google Patents

Description

The present invention relates to a fingerprint pattern matching method, and more particularly to a fingerprint pattern matching method in a fingerprint reading device and a fingerprint matching processing device.

[Conventional technology]

In the conventional fingerprint pattern matching method, for each of the search fingerprint pattern and the file fingerprint pattern to be compared with the search fingerprint pattern, the feature of each coded pattern is compared with the feature point number of the search fingerprint pattern and the feature point number of the file fingerprint pattern. I remember. Then, the data showing the features of each pattern includes feature point number N, feature point type Q, feature point coordinates (X, Y), direction D, congestion amount C, and four divided points centering on the feature point. There are other closest feature point numbers M0 to M3 in the quadrant, and the number of ridges R0 to R3 interposed between the feature point and other closest feature points in each quadrant.

Also, as a feature of the conventional fingerprint pattern matching method,
Include the use of parent, child and grandchild feature points. That is, one feature point serving as a reference is used as a parent feature point, and the nearest point to the parent feature point in each quadrant divided into four with this feature point as a center is set as a child feature point, and the child feature point is divided into four as a center. In addition, the nearest point to the child feature point in each quadrant is the grandchild feature point. Then, a set of these parent, child, and grandchild feature points is considered as one matching unit, and the search fingerprint pattern and the file fingerprint pattern are sequentially compared by these matching units. Therefore, when there are n feature points in each of the search fingerprint pattern and the file fingerprint pattern, these are the parent feature points, and therefore the matching of the parent feature points is performed as follows.
It must be repeated n ² times, which is a very iterative calculation including the matching of child and grandchild feature points.

[Problems to be Solved by the Invention]

The above-mentioned conventional fingerprint pattern matching method has the coordinate information of each feature point in the data of each feature point, so that all the feature points of the file fingerprint pattern can be searched for one of the feature points of the search fingerprint pattern. In addition to the above, there is a problem that the number of searches becomes very large in order to compare the respective feature points of the parent, child, and grandchild.

Furthermore, in the conventional fingerprint pattern matching method that uses the feature points of the parent, child, and grandchild, the number of ridges existing between the parent and the child or grandchild serves as the information of the parent feature point. Since the number of ridges from the parent to the grandchild is large, it is slightly reliable because the number of ridges from the parent to the grandchild is estimated from the number of ridges from the parent to the child and from the child to the grandchild. Has the problem of falling.

From these things, the conventional fingerprint pattern matching method is
Although the matching accuracy is quite high, it has a drawback that it takes a lot of time to perform the repetitive processing.

[Means for solving the problem]

The fingerprint pattern matching method of the present invention is a fingerprint pattern matching method for comparing and matching a search fingerprint pattern to be identified and at least one pre-registered file fingerprint pattern to be matched with the pattern fingerprint information based on the feature point information of each pattern. In the above, means for displaying the position of the feature point of the pattern by polar coordinates whose origin is the center point of the fingerprint to which the feature point belongs, and radiation which is predetermined at equal intervals with the center point of the fingerprint as the origin and some predetermined Means for creating a storage area corresponding one-to-one to each cell divided into blocks by concentric circles by radius, means for storing information of characteristic points in the cells corresponding to each of the storage areas, one cell The number of feature points included in the cell is stored in the storage area corresponding to the case where two or more feature points are included in And the storage area corresponding to the cell is prepared for each of the search fingerprint pattern and the file fingerprint pattern, and the contents of the storage area corresponding to the cells on the same coordinates are respectively compared and calculated by an arithmetic circuit. Means for determining the collation result of the fingerprint of the search fingerprint pattern and the fingerprint of the file fingerprint pattern based on the result of the comparison operation, and the coordinate conversion of the rotational movement in units of at least one cell by one command. And a means for logically synthesizing storage areas corresponding to a predetermined number of cells that are adjacent to each other in coordinates to perform a rough collation and handle them as one large enlarged cell.

〔Example〕

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a fingerprint explanatory view showing an example of a fingerprint pattern. FIG. 2 is a coordinate configuration diagram showing an example of polar coordinates for the fingerprint pattern of FIG. The origin (center point) of the coordinates in FIG. 2 corresponds to the center point of the fingerprint pattern in FIG. 1 or an appropriate position on the center axis when the center point is unknown. Then, the radiation θi-θi-1 at the coordinates in FIG.
The portion surrounded by the concentric circles rj-rj-1 is a cell. The number of radiation, the angle, the diameter of the concentric circles, the number of radiation, etc. are first determined in advance as parameters.

FIG. 3 is an explanatory diagram of cell numbers showing an example of cell numbers in FIG. The cell numbering is as shown in FIG.
They are arranged in a clockwise order from the central part to the outer part.

FIG. 4 is an explanatory diagram of characteristic point coordinates showing an example of characteristic point coordinates in the polar coordinate display of FIG. Further, FIG. 5 is an arrangement relation explanatory view showing an example of arrangement relation between the characteristic points of FIG. 4 and the cells of FIG. As shown in FIGS. 4 and 5,
The characteristic points A, B, C are represented by polar coordinates as shown in FIG. 4, and it is checked in which cell the characteristic points A, B, C are included in FIG. That is, first, the distance from the center of the feature point is checked, and the value of j when the feature point is included between the radii rj and rj−1 is obtained. The j of the characteristic point A shown in FIG. 5 is 2, the j of the characteristic point B is 3, and the j of the characteristic point C is 2. Next, the angle is examined to find the value of k when the feature point is included between the angles θk and θk−1. The k of the characteristic point A shown in FIG. 5 is 2, the k of the characteristic point B is 2, and the k of the characteristic point C is 1.
Is. These two numbers j, k are investigated for all feature points. As a result, the values of j and k are assigned to all the feature points, which makes it possible to assign the feature points to the cells associated with (j, k).

FIG. 6 is an information structure diagram showing an example of each table in which the values of j and k are assigned. FIG. 6 (a) is a correspondence table of cell numbers and j, k, FIG. 6 (b) is a correspondence table of distances r and j from the center, and FIG. 6 (c) is a correspondence table of angles θ and k. The corresponding tables are shown.

Then, the information on the feature points within the coordinates of each cell is stored in the storage area corresponding to that cell.
This is the storage area corresponding to one coordinate cell,
It is called a memory cell.

FIG. 7 is an information structure diagram showing an example of a memory cell corresponding to one cell. FIG. 7 (a) shows an example in which there is one feature point in the cell on the corresponding coordinates, and FIG. 7 (b)
Indicates an example in which there are two or more. E in FIG. 7 is 2
It has a bit, the upper bit indicates the presence or absence of the feature point, "1" when it is present, "0" when it is not present, and the lower bit indicates the number of feature points, and when it is 1 or less "0", 2
It is "1" when there are more than one. Then, the contents of the following memory cells are changed by this lower bit, and the form of FIG. 7 (a) is obtained in the case of "0", and the form of FIG. 7 (b) is obtained in the case of "1". .

FIG. 8 is an information structure diagram showing an example of Q and M in FIG. As shown in FIG. 8A, Q indicates the type of feature point with 2 bits. On the other hand, C in FIG. 7 (b) indicates the number of feature points, and 4 bits can indicate a maximum of 15 in one cell. FIG. 9 is an explanatory diagram showing an example of how to obtain the direction of the ridge. D in FIG. 7 (a) is the direction of the ridge at the feature point, and as shown in FIG.
The counterclockwise angle θ (360 degrees is shown as 256) is expressed based on the clockwise direction of the tangent of the concentric circle from the center point.

Further, V in FIG. 7 (a) indicates the curvature of the ridge, and the direction D
On the other hand, the left tune is represented by + and the right tune is represented by −, and is represented by a numerical value of -7 to 7. On the other hand, R in FIG. 7 (a) is the distance to the closest other feature point and is expressed by 0 ≦ R ≦ 511. Further, M in FIG. 7 (a) is the number of ridges, and indicates the number of ridges existing between other closest feature points by 2 bits, and its content is shown in FIG. 8 (b). ).

FIG. 10 is an information structure diagram showing an example of the relationship between the cell number and the contents of the storage area. The correspondence between the cells on the coordinates and the storage cells is shown in FIG. In addition, all the bits of the cells having no characteristic points are 0.

FIG. 11 is an explanatory diagram of a comparison calculation showing an example of a comparison calculation of storage cells on the search side and the file side. Fig. 12 shows
FIG. 12 is a comparison calculation explanatory diagram showing in detail an example of a comparison calculation of the pair of cells in FIG.

Next, the matching method will be described. Matching has up to the third stage. First, in the first stage, only E, that is, only the coordinate position of the feature point is compared. In the arithmetic circuit E of FIG. 12, whether or not there is a characteristic point of the cell, and if there is,
It is comparing whether it is two or more. The output from the arithmetic circuit E is 2 bits, and the upper bit shows the comparison result of the presence or absence of the feature. The lower bits are not used in the first stage of matching. This collation is performed for all cells by the E arithmetic circuit of each cell. From this result, it is determined whether to perform the next second-stage matching or to perform coordinate conversion depending on how many cells match, whether they match a lot at the center of the fingerprint, and whether they match a lot outside. In the case of coordinate conversion, there is rotational movement or the like, but in any case, coordinate conversion is performed by moving data to an adjacent memory cell on the moving direction side. Then, the first-stage matching is performed again.

Next, in the second-stage matching, Q or C matching is performed. In this collation, only cells that are matched in the first-stage collation are targeted for comparison. Since only the upper bits of E are compared in the first-stage comparison, if the number of feature points in the two cells being compared is one or more, it is considered that they match. Especially when the number of feature points in either cell is two or more, the lower bit of the operation output of E becomes "1", and this signal causes the operation circuit of NO.2 in FIG. It is determined to be a circuit for performing the calculation of.

FIG. 13 is a characteristic point calculation explanatory view showing an example of the comparison of the characteristic points of FIG. 12 and the output thereof. The calculation of C is the collation of the number of feature points, and the output result of the collation is shown in FIG. Then, information about at which position the cell matching result is added is added to this result. Then, the collation result of the central portion of the fingerprint is taken heavy.

Further, when the lower bit of the arithmetic circuit of E is "0", that is, both have one or less characteristic points in the cell, N of FIG.
The operation circuit of O.2 collates the feature point type Q. In this case, the output is 1 bit, which is "1" when they match and "0" when they do not match.

Whether these calculations are performed for each cell, the results are totaled, compared with a predetermined reference value, and whether the third step comparison is performed or whether coordinate conversion is performed depending on whether or not the reference value is reached. To decide. The coordinate conversion is similar to the case of the first stage, and after the coordinate conversion is completed, the process returns to the collation of the first stage again.

Next, the third stage matching is a case where the number of feature points in the cells on both the search side and the file side is one, and D (direction), V
(Curvature), R (distance to nearest feature point), M (number of ridges) are compared and collated.

FIG. 14 is an explanatory diagram of parameter calculation showing an example of a method of expressing each parameter of FIG. 12 and its comparison calculation result. The four parameters are compared with each other. Each parameter can be weighted, and the weight is set according to which parameter is considered important. FIG. 14 shows an example of output contents from each arithmetic circuit. This output is sent to each weighting circuit, and the output results are totaled to calculate a score.

Note that there may be two or more candidates that were issued from the final stage. In this case, the scores are used for ranking.

Next, regarding the coordinate conversion of the fingerprint pattern, although the coordinate conversion must be calculated in the conventional fingerprint pattern matching method, in the present embodiment, the storage cells are connected to facilitate the conversion beforehand.

FIG. 15 is a block diagram showing an example of the coordinate conversion circuit. 16 is a block diagram showing an example of a circuit for one memory cell shown in FIG. As shown in Fig. 16, the input is a data input (DIN) for loading data from a data file or input / output device, and a shift input (SIN) for loading data that is shifted from an adjacent cell by coordinate conversion. have. Output is data output (D
OUT), which also serves as the shift output. These input / output control is performed by a data input / output instruction (DC) and a data shift instruction (DSFT).

For example, when the clockwise rotational movement is performed, the direction of the bidirectional multiplexer shown in FIG. 15 is first determined, and the clockwise rotational movement inputs the data from A and outputs the data output to B. Connected to do. Then, the data from the A is input to the cell with the cell number n by the data shift instruction, and the stored data is output to the B at the same time. By these operations, the rotational movement for one cell is performed.

FIG. 17 is a coordinate configuration diagram showing an example of one enlarged cell obtained by combining four cells. In this embodiment, a method as shown in FIG. 17 can be adopted in order to make the matching accuracy coarse. That is, four adjacent cells are combined and treated as one expanded cell. By performing such a method on the file side, it is processed as if it had a rough cell structure.

FIG. 18 is a block diagram showing an example of a circuit for forming an enlarged cell by a precision changing instruction. The circuit shown in FIG.
Installed in the part of the figure. In this circuit, the switch circuit is switched to the synthesizing circuit side by the precision changing instruction.
In the synthesizing circuit, if there is no feature point in the four memory cells, "0" is substituted in the memory of the expanded cell. In the case of only one cell, the information of the existing cell becomes the information of the expanded cell as it is. On the other hand, in the case of two or more, the number of feature points is stored and "0" is input to the remaining D, V, R and M.

〔The invention's effect〕

As described above, the fingerprint pattern matching method of the present invention uses the memory cells corresponding to the cells surrounded by the radiation and the concentric circles on the fingerprint pattern, so that the cells become larger toward the peripheral portion. In addition, the error due to the distortion, which becomes more prominent toward the periphery of the fingerprint pattern, can be reduced, and all memory cells can be operated in parallel, which has the effect of significantly improving the operation speed. ing.

Further, in the fingerprint pattern matching method of the present invention, when the center points of both the search fingerprint pattern and the file fingerprint pattern are known, coordinate conversion can be performed only by rotational conversion, which is advantageous for polar coordinates, so that particularly high speed processing is possible. Has the effect of being

As a result, the present invention can be expected to exert a tremendous effect on the processing of an imprinting type fingerprint having a clear fingerprint pattern.

[Brief description of drawings]

FIG. 1 is an explanatory view of a fingerprint showing an example of a fingerprint pattern, FIG. 2 is a coordinate configuration diagram showing an example of polar coordinates for the fingerprint pattern of FIG. 1, and FIG. 3 is a cell showing an example of cell numbers in FIG. FIG. 4 is an explanatory diagram of characteristic point coordinates showing an example of characteristic point coordinates in polar coordinate display of FIG. 2, and FIG. 5 is an example of an arrangement relationship between characteristic points of FIG. 4 and cells of FIG. FIG. 6 is an information configuration diagram showing an example of each table in which the values of j and k are assigned, FIG. 7 is an information configuration diagram showing an example of a memory cell corresponding to one cell, FIG. 8 is an information structure diagram showing an example of Q and M in FIG. 7, FIG. 9 is an explanatory diagram showing an example of how to obtain the direction of the ridge, and FIG. 10 is a relation between the cell number and the contents of the storage area. FIG. 11 is an information configuration diagram showing an example, and FIG. 11 is an explanatory diagram of a comparison operation detailing an example of a comparison operation of the memory cells on the search side and the file side FIG. 12 is a comparison calculation explanatory view showing an example of the comparison calculation of the pair of cells in FIG. 11 in detail, FIG.
FIG. 12 is a diagram for explaining the calculation of the characteristic points showing an example of the comparison of the characteristic points of FIG. 12 and its output, and FIG. 14 is the calculation of the respective parameters of FIG. 12 and an example of the expression method of the comparison arithmetic result. Explanatory diagram, FIG. 15 is a block diagram showing an example of a coordinate conversion circuit, FIG. 16 is a block explanatory diagram showing an example of a circuit for one memory cell of FIG. 15, and FIG. 17 is a composite of four cells. A coordinate configuration diagram showing an example of one enlarged cell,
FIG. 18 is a block diagram showing an example of a circuit for forming an enlarged cell according to a precision change command.

Claims (1)

(57) [Claims] 1. A fingerprint pattern matching method in which a search fingerprint pattern to be evaluated and at least one pre-registered file fingerprint pattern to be matched with the fingerprint pattern are compared and matched according to the feature point information of each pattern. By means for displaying the position of the characteristic point in polar coordinates with the center point of the fingerprint to which the characteristic point belongs as the origin, and with the radiation center and the concentric circles having some predetermined radii determined in advance with the center point of the fingerprint as the origin. Means for creating a storage area corresponding to each cell divided into one to one, means for storing information of feature points in the cell corresponding to each of the storage areas, two in one cell A means for storing the number of feature points included in the cell in the storage area corresponding to the case where the feature points are included; Means for preparing the storage area for each of the search fingerprint pattern and the file fingerprint pattern, and performing a comparison operation on the contents of the storage area corresponding to the cells on the same coordinates, respectively; Rough matching is performed with a unit that determines the matching result of the fingerprints of the search fingerprint pattern and the file fingerprint pattern based on the result of the operation, a unit that performs coordinate conversion of the rotational movement in units of at least one cell by one command. Therefore, the fingerprint pattern matching method is characterized by further comprising means for logically synthesizing storage areas corresponding to a predetermined number of cells adjacent on the coordinate and handling them as one large enlarged cell.