JPH0654186A - Image recognizing device - Google Patents

Abstract

PURPOSE:To certainly prevent the forgery of paper money, a valuable securities, etc. by comparing an inputted digital image and a stamp pattern previously- registered in a memory, judging it to be a stamp when they are coincident and discriminating a picture peripheral to the stamp to judge it whether to be paper money or not. CONSTITUTION:A digital color image input part 101 reads paper money, converts the piece of image information to the digital color signal of eight-bits of respective colors R, G and B and outputs it to a pattern comparing part 103. The comparing part 103 reads the image pattern of the stamp of paper money which is previously registered in a pattern register part 102, compares it with the input color image and judges whether it is the stamp of paper money or not. At the time of judging it to be the stamp of paper money, a peripheral data discriminating part 104 extracts the stamp of the input color image and its periphery next, judges the presence/absence of the picture and judges it to be paper money at the time of presence. Thus, even when there is a scribble, etc., paper money, the valuable securities, etc., are accurately discriminated and the forgery is certainly prevented with simple constitution.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image recognition apparatus for recognizing an input digital image, and more specifically, comparing the input image with a pattern image registered in advance and extracting the pattern image. The present invention relates to an image recognition device for recognizing special documents such as banknotes and securities by determining the state of the peripheral area in the image portion.

[0002]

2. Description of the Related Art As a conventional image processing apparatus for preventing forgery of banknotes and the like, there is an "image processing apparatus" disclosed in Japanese Patent Application Laid-Open No. 2-210591. This image processing apparatus includes a detection unit for detecting the position state of an original document, and a determination unit for extracting image data of a specific portion in the original document according to the detection by the detection unit and determining the degree of similarity with a pattern registered in advance. And a specific document is discriminated based on the similarity.

[0003]

However, in the conventional image processing apparatus as described above, since it is determined whether or not it is a bill based on the degree of similarity with a specific pattern registered in advance. , The configuration for improving the accuracy of pattern recognition becomes complicated (for example, the pattern recognition accuracy cannot be improved unless there is a large amount of information about the pattern to be registered in the memory in advance because it is determined only by the similarity). However, when there is noise information such as graffiti on the bill, there is a problem that misjudgment often occurs.

The present invention has been made in view of the above, and accurately discriminates the type of an original with a simple structure, and ensures the forgery of banknotes, securities, etc. using an image forming apparatus such as a copying machine. The purpose is to prevent.

[0005]

In order to achieve the above-mentioned object, the present invention provides an image input means for inputting a digital image, a pattern registration means for registering a predetermined pattern image, and the image input means. Based on the state of the peripheral area of the image portion extracted by the comparing means, comparing the digital image input from the above with the pattern image registered in the pattern registering means and extracting the compared portion. (EN) An image recognition device provided with a discrimination means for discriminating whether a document is a special document.

Further, it is desirable that the special manuscript is a bill or a securities.

Further, it is desirable that the pattern image registered in the pattern registration means is a round outer shape of a seal on a bill, and the comparison means compares the shape and color with the input image data.

Further, it is preferable that the discriminating means discriminates whether the peripheral area of the extracted image portion is a pattern or not, and discriminates it as a special original when it is judged to be a pattern.

[0009]

The image recognition apparatus according to the present invention inputs a digital color image, compares the digital color image with, for example, a pattern image relating to a round outer shape of a stamp of a banknote registered in a memory, and regarding its shape and color. It is judged whether or not they match each other, and if they match and it is judged that they are the above-mentioned seals, the seal and its peripheral portion in the input digital color image are extracted and it is judged whether or not a pattern exists in the peripheral portion. Determine whether. As a result, when it is determined that the pattern exists, it is determined that the document is a bill.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a block diagram showing a schematic configuration of an image recognition device according to the present invention. In the figure, 101 is a digital color image input unit configured by a color scanner, for example, for reading image information of banknotes, R,
Input an 8-bit digital color image for each of G and B.
Reference numeral 102 denotes a pattern registration unit in which an image pattern as shown in FIG. 4 (for example, a round outer shape of a stamp of a bill) is stored in advance, and is composed of a ROM (Read Only Memory). Further, the pattern registration unit 102 limits the colors of the seal to the following distributions. R (red) = 130 to 140 G (green) = 10 to 20 B (blue) = 5 to 15

Reference numeral 103 denotes a pattern comparison unit that compares the digital color image input from the digital color image input unit 101 with the pattern data registered in the pattern registration unit 102 and extracts the image. Only when the pattern data and the pattern data match each other in shape and color, it is judged as a stamp of a banknote. Also, 1
Reference numeral 04 is a peripheral data discriminating unit that extracts the periphery of the seals compared by the pattern comparing unit 103 and discriminates whether or not a pattern exists based on a difference between a plurality of peripheral data and an average value of the peripheral data.

The operation of the above configuration will be described with reference to the flow chart shown in FIG. The image recognition operation of 1,000 yen (hereinafter referred to as 1,000 yen bill) of the Bank of Japan bill shown in FIG. 3 will be described as an example. First, the digital color image input unit 101 reads the image information of the 1,000-yen bill, converts the image information of the 1,000-yen bill into an 8-bit digital color signal for each of R, G, and B, and inputs it to the pattern comparison unit 103 ( S20
1). In the pattern comparison unit 103, the pattern registration unit 10
The image pattern (round outer shape of the stamp of the bill) as shown in FIG. 4 which is registered in advance in 2 is read out and compared with the digital color image of the 1,000-yen bill input through the digital color image input unit 101, It is determined whether the digital color image can be recognized as a stamp of a bill (S202).

As a result, when it is determined that the digital color image can be recognized as a stamp of a banknote, the peripheral data discriminating unit 104 then outputs the digital data extracted by the pattern comparing unit 103 to the state shown in FIG. The seal of the color image and its periphery are extracted, and it is determined whether or not a pattern exists in the periphery (S203). This peripheral data discrimination unit 10
Regarding the judgment as to whether or not there is a picture in the periphery by the method 4, the sampled peripheral data A, B,
In C and D (about 10 pixels each), when the difference from the average value is below a certain level, the peripheral data is recognized as having no pattern.

FIG. 6 is a block diagram showing a schematic configuration of the peripheral data discriminating section 104. The peripheral data (A, B, C,
D) 601 and the average value 602 obtained by calculating the average of the peripheral data 601 are input to the A terminal and the B terminal of the comparator 603, respectively, and if | AB | is less than a certain numerical value (for example, 10), The data is judged to have no pattern (uniform density). This operation will be described with reference to the flowchart of FIG. First, the peripheral data 601 is transferred to the peripheral data discriminating unit 10
4 (S701), and based on the comparison result of the comparator 603, it is determined whether the difference from the average value is 10 or less (S702). Here, when it is determined that it is 10 or less, it is determined that there is no pattern (see FIG. 8A), and conversely,
When it is determined that the number is 10 or more, it is determined to be a pattern (see FIG. 8A).

As a result, when it is determined that it is a design, it can be determined that it is a bill (S704). This is because a normal seal is stamped on the base part, but a stamp on a banknote is stamped on the design. In this way, since the double check of the pattern check in the pattern comparison unit 103 and the state check of the peripheral portion in the peripheral data determination unit 104 is executed, even when there is noise information such as graffiti on the banknote, Misjudgment can be reliably prevented, and bills can be reliably recognized with a simple configuration. The flowcharts shown in FIGS. 2 and 7 above are
It is a part (subroutine) in multiple flows, and this flow does not mean that the entire series of operations ends.
After the operation according to the above flow is completed, the process shifts to a flowchart for another operation.

[0016]

As described above, according to the image recognition apparatus of the present invention, a digital color image is input, and the digital color image and, for example, a pattern relating to the round outer shape of the stamp of the banknote registered in the memory are input. If they are compared, it is judged whether or not they match each other in terms of shape and color, and if they match and it is judged that they are the above seals,
The periphery of the seal is extracted and it is determined whether or not there is a design. As a result, when it is determined that the pattern is present, it is determined that the original is a banknote. Therefore, the type of the original is accurately determined by a simple configuration, and the banknote and the valuable note using an image forming apparatus such as a copying machine are used. Forgery of securities etc. can be reliably prevented.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of an image recognition apparatus according to the present invention.

FIG. 2 is a flowchart showing an operation of the image recognition device shown in FIG.

FIG. 3 is an explanatory diagram showing a 1,000-yen bill of a Japanese banknote as an example of a document input by a digital color image input unit according to the present invention.

FIG. 4 is an explanatory diagram showing an example of an image pattern registered by a pattern registration unit according to the present invention.

FIG. 5 is an explanatory diagram showing a seal of a digital color image extracted by a pattern comparison unit according to the present invention and its peripheral portion.

FIG. 6 is a block diagram showing a schematic configuration of a peripheral data discriminating unit according to the present invention.

FIG. 7 is a flowchart showing an operation of a peripheral data discriminating unit according to the present invention.

FIG. 8 is a graph showing a pattern recognition standard and a pattern recognition standard according to the present invention.

[Explanation of symbols]

101 digital color image input unit 102 pattern registration unit 103 pattern comparison unit 104 peripheral data discrimination unit

Claims (4)

[Claims] 1. An image input means for inputting a digital image, a pattern registration means for registering a predetermined pattern image, a digital image input from the image input means, and a pattern registered for the pattern registration means. Comparing images, comparing means for extracting the compared parts, and judging means for judging whether or not it is a special document based on the state of the peripheral area of the image parts extracted by the comparing means. An image recognition device characterized by. 2. The image recognition apparatus according to claim 1, wherein the special document is a bill or a securities. 3. The pattern image registered in the pattern registration means is a round outer shape of a seal on a banknote, and the comparison means compares the shape and color with the input image data. The image recognition device according to item 1. 4. The image according to claim 1, wherein the determining means determines whether or not the peripheral area of the extracted image portion is a pattern, and when it is determined that the pattern is a pattern, the image is a special document. Recognition device.