JPH0121385Y2 - - Google Patents

Description

[Detailed explanation of the invention] A. Industrial application field This invention is based on the invention, in which the impression of the seal that a customer submits when requesting payment of a deposit at the counter of a financial institution such as a bank is the same as the impression of the customer's registered seal. This relates to a device that automatically verifies whether the

B Summary of Disclosure The automatic seal stamp verification device of the present invention stores in advance the XY coordinate image of the seal impression of a registered seal impression (hereinafter referred to as the registered seal impression) as well as the location of the characteristic area in the polar coordinate image of the seal impression, and allows the customer to make a payment request. The polar coordinate image of the seal impression of the seal to be submitted at the time (hereinafter referred to as the verification seal impression) is obtained, and the verification seal impression is created so that the correlation value between the above characteristic region and the portion corresponding to the above characteristic region in the polar coordinate image of the verification seal impression is maximized. The above rotation is performed by correcting the origin of the Kyokuzen mark, detecting the rotation angle with respect to the registered seal imprint, and then inversely transforming the polar coordinate image of the verification seal imprint obtained based on the finally corrected origin into an XY coordinate image. By rotating only the corner, the verification seal imprint is aligned with the registered seal imprint, and the matching rate between them is calculated. If this matching rate is greater than or equal to a predetermined value, it is determined that the verification seal imprint and the registered seal imprint are the same. The present invention is characterized by providing means for accurately correcting the center position of the verification seal impression and detecting the rotation angle with respect to the registered seal impression using a polar coordinate image of the seal impression.

C. Prior Art JP-A-56-157579 discloses a system in which a registered seal impression is represented by a plurality of concentric circular patterns, and a specific concentric circular pattern selected based on their autocorrelation values is registered in a file, and obtained from the verified seal imprint. By calculating the rotation angle deviation between the registered seal imprint and the verification seal imprint based on the correlation value between each concentric circle pattern and the registered specific concentric circle pattern, and aligning one seal imprint by rotating one of the seal imprints by that angle. The present invention discloses a device that verifies these seal impressions. In matching such circular patterns, it is necessary to accurately determine the center position of the circle, but conventionally, the center position has been determined by measuring the maximum vertical and horizontal widths of the pattern. It was common.

D. Problems that the invention aims to solve With the conventional technology as described above, if there is "fading" or dirt on the frame of the seal, an error occurs in measuring the center position, so it is difficult to obtain accurate verification. I couldn't do it.

E. Means for Solving the Problem (FIG. 1) In the present invention, a center position correction section is provided in order to determine the accurate center position of the verification seal imprint. This center position correction unit uses the center position P ₀ obtained by the conventional technique as described above as a temporary center position, and calculates the center position from that P ₀ as a temporary center position.
Select points P ₁ to _{P 8} that are one pel apart on the XY coordinates, and compare the characteristic areas in the registered seal impression with the parts corresponding to the above characteristic areas in each polar coordinate image of the verification seal impression with P ₀ to P ₈ as the coordinate origin. Find the correlation value between. Among the polar coordinate images with P ₀ to _{P 8} as the origin, select the one showing the maximum correlation value, for example P ₆ , and use it as the temporary center position, and then calculate again for multiple points l/2 pels away from that position. Perform the above operations. In this way, by repeating this operation an arbitrary predetermined number of times, the accurate center position of the verification seal impression can be obtained.

F. Embodiment The seal verification device of the present invention is composed of a plurality of blocks as shown in FIG. 2, and the function of each block will be explained according to its operation steps.

CCD scanner 1 has an impression field mechanism and CCD
The scanner quantizes the scanned image of the seal imprint into 16-value levels and sequentially transfers it to the image buffer 2.

The image buffer 2 is a 1 megabyte memory that can be shared by the scanner 1, polar coordinate converter 5, correlator 7, coordinate inverse converter 9, and a main controller (not shown) that controls the entire device. It has space.

The binarization unit 3 is the 16 in the image buffer 2.
A binarized image is obtained from a value-level seal imprint image. When registering a seal impression, the optimum binarization level is determined by manual operation while displaying the seal impression on a display device. If you want to binarize the verification seal impression,
The optimum binarization level is automatically determined based on the total number of black dots of the registered seal impression to obtain a binarized image.

The center position detection section 4 determines the center position of the seal impression by detecting the upper, lower, left and right edges of the seal impression. The center position obtained by this method tends to vary even if the seal is the same, as the frame of the seal impression is blurred or missing when the seal is stamped, and therefore the detected center position needs to be corrected. .

The polar coordinate converter 5 converts the image into a polar coordinate image and stores it in the image buffer 2 by specifying the center position and conversion area of the XY coordinate image of the seal imprint in the image buffer 2. FIG. 3 shows the state of this conversion. The transformation area is specified by R1 to R16. Each area is
It has a width of 16 pels and a length of 2048 pels.

The characteristic region determination unit 6 determines the characteristic region of the registered seal imprint to be used for center position correction and rotation angle detection by Ri (i=
Decide from 1 to 16). To make this determination, find the autocorrelation function in the angle, that is, the θ direction, for each region of R1 to R16, and calculate the region where the integral value S near τ=O (−kτk) is the minimum among the regions R1 to R16. Let be the feature region Rp. The autocorrelation function ΦRi of the region Ri on the polar coordinates is defined as follows.

ΦRi(τ) = ₂₀₄₇ 〓 〓 ⁼⁰ _16(i-1)+15 〓 〓 ^=16(i-1) Ri(r, θ) Ri(r, θ−τ) However, Ri(r, θ) is 1 or 0 as a periodic function of 2π
Take the two values of △θ=2π/2048, △r=1/16 (mm)
shall be. or represents an exclusive NOR.

The area where the integral value S is minimum is determined by the following equation.

S(Rp) = min ^{1 i 16} _k 〓 〓 ^=k ΦRi(τ) In other words, the autocorrelation function in the region R ₁ to R ₁₆ is τ =
A region where the steepest peak is obtained near 0 is selected as a feature region. Note that k is given as an arbitrary constant. The area determined in this way is stored as a characteristic area of the registered seal imprint when the seal is registered.

Correlator 7 compares two regions in image buffer 2 and calculates the number of black pels in each region and the number of matched black pels in those two regions. The size of each region is 16 pels x 2048 pels. As shown in FIG. 4, this comparison can be made with one region moved by τ pels (0τ2047). This correlator can perform the calculation of an autocorrelation function by the feature region determination section 6, the calculation of a cross-correlation function by the center position correction section 8, and the calculation of the arrival rate by the arrival rate calculation section 10 at high speed.

The center position correction unit 8 compares the characteristic area of the registered seal imprint and its corresponding area in the verification seal imprint using a correlator, and determines the maximum correlation value and the angle τ at that time. This comparison is made as shown in FIG. That is, (a) Let the center position of the verification seal impression obtained by the center position detection section 4 be _P0 , and select positions _P1 to _P8 that are one pel apart on the XY coordinates.

(b) Convert the verification seal impression into a polar coordinate image with P ₀ to _{P 8} as the coordinate origin, compare the characteristic region of the registered seal impression and the corresponding region of each polar coordinate image using the correlator 7, and compare each region with the characteristic region. Find the maximum correlation value between.

(c) Select the origin position and rotation angle of the polar coordinate image that shows the highest correlation value among those maximum correlation values, and set that origin position as a new P ₀ (b)
, where new positions P ₁ to P ₈ are chosen to be l/2 pels away from P ₀ .

(d) By repeating steps (b) and (c) above, the center position of the verified seal impression can finally be made to coincide with the center position of the registered seal impression.

When the polar coordinate image and rotation angle τ in the image buffer 2 are specified, the coordinate inverse transformer 9 rotates the polar coordinate image by τ and creates an XY coordinate image of the obtained image on the image buffer. That is, a polar coordinate image of the verification seal imprint whose coordinate origin is the center position finally determined by the center position correction unit 8 is created by the polar coordinate converter 5, and the maximum correlation value obtained by the center position correction unit 8 is By inversely transforming the coordinates based on the given rotation angle τ, it is possible to obtain a verification seal imprint that is aligned with the registered seal imprint on the XY coordinates.

The matching rate calculation unit 10 uses the correlator 5 to calculate the matching rate between the registered seal imprint and the aligned verification seal imprint. The matching rate M is defined as follows.

M=2Hb/Rb+Sb×100 (%) where Hb is the number of matched black pels, Rb is the total number of black pels in the registered seal impression, Sb is the total number of black pels in the verification seal impression.The match rate obtained in this way is the specified number. When the value is larger than the value of , the verification seal impression is determined to be the same as the registered seal impression. In addition, CCD scanner 1,
The image buffer 2, the binarization section 3, the center position detection section 4, and the polar coordinate converter 5 can be used in common for processing both the registered seal impression and the verification seal impression.
Of course, they may be provided separately for each process.

G. Effect It is possible to align the verification seal imprint with respect to the registered seal imprint and to calculate the matching rate accurately and at high speed.

[Brief explanation of drawings]

Figure 1 is a diagram conceptually showing a method for correcting the center position of a verification seal impression, Figure 2 is a block diagram of the invented seal verification device, and Figure 3 is an XY coordinate image of the seal impression and its corresponding polar coordinate image. A diagram showing
and FIG. 4 conceptually illustrate a method for comparing two regions in a correlator.

Claims (1)

[Claims for Utility Model Registration] Means for optically scanning a registered seal impression to generate a multi-level quantized signal; means for binarizing the quantized signal representing the registered seal impression; of the registered seal impression represented by the binary coded signal;
a first polar coordinate conversion means for converting an XY coordinate image into a polar coordinate image; a means for determining and storing a characteristic region in the polar coordinate image; and an optical scanning means for optically scanning a verification seal impression to generate a multi-level quantized signal. means for binarizing the quantized signal representing the verification seal impression; means for detecting the center position of the verification seal impression; and means for detecting the center position of the verification seal impression represented by the binary signal.
a second polar coordinate conversion means for converting the XY coordinate image into a polar coordinate image having the center position as the origin; an image obtained by digitally rotating the polar coordinate image of the verification seal impression; a correlation means for obtaining a correlation value between the registered seal imprint and the characteristic region in the polar coordinate image; center position correction means for supplying to the polar coordinate conversion means, and the registration by inversely converting the polar coordinate image of the verification seal imprint into an XY coordinate image at a rotation angle that gives the maximum correlation value finally obtained in the correlation means. means for obtaining a verification seal imprint that is aligned with respect to the XY coordinates of the seal impression; and means for obtaining a matching rate between the aligned verification seal impression and the registered seal imprint, and the center position correction means is configured to adjust the new center An automatic seal verification device characterized by repeating position supply a predetermined number of times.