JPS5951026B2 - Image matching device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image matching device that determines whether two images are the same image, and in particular, the present invention relates to an image matching device that determines whether two images are the same image. It relates to a device for accurate verification.

A pattern matching method using digital processing is often used to compare and match the patterns of seal impressions, characters, printed matter, etc., with the patterns on the original register.

Figure 1 a and b are the binarized patterns of the seal frame,
On the two-dimensional coordinates of Xy, a point with a signal is logical
A point with no signal is represented by a logic "O", and a black circle in the diagram indicates that a logic "1" bit exists in this block. When the two patterns shown in FIG. 1A and b are superimposed and compared, the result shown in FIG. 1c is obtained. Here, the Tomari mark is a point where the patterns overlap, and the ゛△''゛×'' mark is a point that is on one side but not on the other. The following is often used as the degree of matching F between these two patterns.

However, S.

is the area of the overlapping part of the two patterns (the total number of white circles in Figure 1c), S,,S. are the areas of each of the two patterns (the total number of black circles in FIG. 1A and b).
By the way, this kind of verification is generally done in a file with the original pattern, and is done to determine whether the input pattern matches the original pattern.
Input patterns generally include various types of noise.

In particular, patterns using ink, vermilion, etc. bleed and become blurred when printed, which results in a lower degree of matching during matching. Even when checking seals, parts of the seal often become blurred or smeared due to the pressure applied when stamping or the way the ink is applied. In addition, even patterns with slight local differences need to be judged as different, but since the above formula (1) considers the degree of matching as the overall average, some patterns may have significant differences. However, if the other parts match well, they may be judged to be the same, which is a drawback. This invention was made in order to eliminate the above-mentioned drawbacks of the conventional devices, and its purpose is to provide a pattern matching device that accurately calculates the degree of matching by taking into account differences in pattern quality and local differences. do.

Embodiments of the invention will be described below with reference to the drawings.

FIG. 2 is a block diagram showing one embodiment of the present invention, in which 1 is a first pattern memory, 2 is a second pattern memory, and 3 is a block diagram showing an embodiment of the present invention.
is the first grayscale memory, 4 is the second grayscale memory, and 5 is the first grayscale memory.
6 is a second feature amount calculation processor, 7 is a data memory, and 8 is a comparison determination processor. Next, the operation will be explained. Pattern memory 1, 2
All bits arranged along the x-y coordinate axes in pattern memories 1 and 2 are compared, for example, by 16 x 16 bits in the x and y directions. First, the characteristic quantity for each area is calculated. The first feature quantity calculation processor 5 calculates the sum of the areas of the parts that do not overlap when the patterns in the pattern memories 1 and 2 are overlapped, as a feature quantity that indicates whether there is a difference between the entire pattern and the patterns in each region. (In other words, the total number of bits that correspond to each other in patterns 1 and 2 and are not a single hole) is calculated as the total area of pattern 1 (that is, the total number of bits in patterns
(total number of bits of logical "l") is calculated and stored in the data memory 7 as the first characteristic quantity.

For each of the above regions, the second feature amount calculation processor 6
As a feature amount that affects the first feature amount in the area,
For example, the density data is read out from the density memories 3 and 4, the average value of the density is determined for the area including the points having density equal to or higher than a predetermined value, and this is set as the second characteristic quantity in the data memory 7.
Enter.

This can be used as a special quantity indicating whether the pattern 1 or 2 is blurred or blurred, and each area can be coded into stages such as blurred, blurred, normal, etc. using this value. The first feature amount and the second feature amount are stored in the data memory 7 for each region, and the matching determination processor 8 uses the first feature amount and the second feature amount to determine the feature of each region. After coding, the entire comparison is judged.

For example, corresponding to the conventional equation (1), ΣF=(L,M)
d (L, M)×α{C (L, M)} ・・・・・・
... Calculate the matching degree F by (2).

However, in equation (2), L and M represent the order of the above regions in the X-axis direction and the y-axis direction, and d(L,.M) represents the order (i.e. coordinate position) in the region of (L, M). The first feature quantity, c(L,M), is a code indicating the second feature quantity in the area whose order is (L,M), α {c(L,M)} is the code c(L,M) The corresponding numerical value Σ (L,M) indicates the sum over all regions.

In the embodiments described above, the average value of the density within each region is used as the second characteristic amount, but instead of this, the dispersion of the density within each region may be used.

That is, in general, if the density is uniform, the image is well-organized, and if the dispersion is large, it can be determined that there is a large amount of dirt, etc., and that differences in patterns are caused by the dirt. In the embodiment shown in FIG. 2, two shading memories 3 and 4 are provided, but if the shading of the original pattern (assuming that it is stored in the pattern memory 1 river) is normal, then , if one of the shading memories is used, the shading memory 3) can be omitted.

Furthermore, it is possible to separate and use the boundary area between the pattern and the background and the other areas out of the areas in the pattern memories 1 and 2 where the patterns do not overlap as the first feature amount or the second feature amount.

Figure 3 is a diagram showing areas where two patterns match and do not match, where ゝ○7 marks are areas where patterns match,
The mark indicates a boundary area where the pattern does not match, and the ゞ×7 mark indicates an area where the pattern does not match but is outside the boundary. The ゞ×7 portion appears only when the pattern difference is large. Since slight deviations and differences in shape appear only in the portion marked with ゞ87, the existence of the portion marked with ゞ×7 is a major factor in determining the difference in patterns. As described above, according to the present invention, a pattern is divided into regions, and for each region corresponding to each other, a first feature amount representing the difference between two patterns and a state that influences this first feature amount are determined. Since the second characteristic amount representing the pattern is determined and comprehensive judgment is made based on this, it is possible to distinguish between differences due to poor pattern quality and minute differences, and perform highly accurate matching. Verification is possible even in certain cases.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an example of pattern matching, FIG. 2 is a block diagram showing an embodiment of the present invention, and FIG. 3 is a diagram showing areas where two patterns match and do not match. 1...First pattern memory, 2...
Second pattern memory, 3...First gradation memory, 4...Second gradation memory, 5...
First feature quantity calculation processor, 6...Second feature quantity calculation processor, 7...Data memory, 8
...Verification judgment processor.

Claims (1)

[Scope of Claims] 1. A first pattern memory that stores a binarized pattern of a first image of two images to be compared with each other, and a first pattern memory that stores a binarized pattern of a first image of two images to be compared with each other; A second pattern memory stores the binarized pattern, and the contents of this second pattern memory are compared with the contents of the first pattern memory, and it is determined that the corresponding contents are the same or mutually compatible. means for calculating a first feature amount for each predetermined area for determining whether there is a difference in content corresponding to the area;
means for calculating a second feature amount that affects the first feature amount for each of the predetermined regions; An image matching device comprising means for determining a result of matching with an image of No. 2. 2. The means for calculating the first feature amount totals the number of bits in which the logic of each bit of the first pattern memory and the logic of each bit of the corresponding second pattern memory do not match within the region. Then, the ratio of this total to the number of bits of logic indicating that it is a figure part in the area of the first pattern memory and/or the second pattern memory is calculated and the first feature quantity is calculated. An image matching device according to claim 1, characterized in that the image matching device calculates . 3. The means for calculating the second feature amount calculates a bit pattern of bits in which the logic of each bit of the first pattern memory and the logic of each bit of the corresponding second pattern memory do not match within the area. , and calculates the second feature quantity by calculating the ratio between the total number of bits on the boundary line with the outside and the total number of remaining bits in this bit pattern. The image matching device according to item 1. 4. Claim 1, wherein the means for calculating the second characteristic amount calculates an average concentration value within the region, and calculates the second characteristic amount as a function of this average concentration value. Image matching device described in section. 5. Claims characterized in that the means for calculating the second feature amount calculates the variance of the density values within the region, and calculates the second feature amount as a function of the variance of the density values. The image matching device according to item 1.