JP2003256837A - Pattern matching processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To quickly judge whether or not an image pattern in a two-dimensional image is matched with a reference image pattern without using any frame memory, and to judge whether or not the both patterns are matched with each other based on the two-dimensional image obtained with a small time interval which is close to a real time. <P>SOLUTION: The image signal of a two-dimensional image obtained by an image pickup part 1 is inputted through an image processing part 2 to a start position/stop position detecting part 9, and the start position/stop position on each raster scanning line of each image pattern in the two-dimensional image is detected. The matching of the start position/stop position with the start position/stop position of the reference pattern is judged by a comparing/judging part 10, and it is judged that the same pattern as the reference image pattern is present in the two-dimensional image according to the matching rate, and the result is announced. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern matching processing method, and more particularly, to a pattern capable of determining the difference between an image pattern in a two-dimensional image and a reference image pattern at high speed without using a frame memory. The present invention relates to a matching processing method.

[0002]

2. Description of the Related Art Pattern matching for determining whether or not an image having a specific pattern exists in a two-dimensional image is used in various fields. For example, in a production line of a factory, pattern matching processing is performed between an external pattern of a component and a reference image pattern, and an inspection device for determining the quality of the component using the result, an automation device for automating mounting and assembly of the component, a robot Has been realized, and an object with a specific pattern is identified by pattern matching processing to automatically track the object and an automatic tracking device that detects the presence or intrusion of dangerous substances by pattern matching processing. Notification devices and the like are also known.

FIG. 9 is a block diagram of an apparatus to which a conventional pattern matching processing method is applied. An image signal of a two-dimensional image obtained by picking up an image of a certain area by the image pickup unit 1 is imaged by the image processing unit 2. After processing such as emphasis and binarization, one screen is written in the frame memory 3. The reference memory 4 stores a reference image signal of a reference image to be compared / determined.

The comparison / determination unit 5 compares the binarized image signal written in the frame memory 3 with the reference image signal stored in the reference memory 4, and based on the comparison result, the two-dimensional image is displayed. It is determined whether or not the reference image exists.
The reference image signal of the reference memory 4 can be obtained by capturing the reference image by the image capturing unit 1.

FIG. 10 is a detailed block diagram of the frame memory 3 and its peripheral circuits. For example, the 8-bit grayscale image signal per pixel is compared with the threshold value TH in the image processing unit 2 including the comparator 6, and becomes a 1-bit binary image signal per pixel. The effective pixels for one screen of the binarized image signal, for example 512 × 512 pixels, are written in the frame memory 3. This writing is controlled by the image transfer clock
The write timing control circuit 7 receives PCLK, the horizontal synchronizing signal HSYN, and the vertical synchronizing signal VSYN as input. In the above example, 512 lines for each 512 lines excluding the horizontal blanking period and the vertical blanking period are used.
Write signal WT to the frame memory 3 in a period of 2 pixels
And the write address signal WADR becomes valid.

The MPU 8 as the comparison / determination unit 5 simultaneously reads the binarized image signals from the frame memory 3 and the reference memory 3 based on the vertical blanking signal VB, and executes the comparison / determination processing.

FIG. 11 is an explanatory diagram of the comparison / judgment processing. Here, as an example, the image signal of the two-dimensional image of 512 pixels in the vertical direction and 512 pixels in the horizontal direction shown in FIG. The reference memory 4 is shown in FIG.
It is assumed that the image signal of the reference image of 128 pixels vertically and 128 pixels horizontally is stored. Further, FIG. 7C shows a state of comparison between the two-dimensional image and the reference image. In addition,
In this example, it is assumed that interlaced scanning is not performed.

In the comparison / determination unit 5, the reference image is sequentially moved relative to the two-dimensional image written in the frame memory 3 one pixel at a time in the horizontal direction from the upper left end to the right end. This is performed by checking the degree of overlap between both images while moving further downward. That is, the MPU 8 outputs the read address signal RADR when the vertical blank signal VB is input, and simultaneously reads the binarized image signal written in the frame memory 3 and the binarized image signal of the reference memory 4 to positionally read the signal. Is compared for each pixel, the matching rate in 128 × 128 pixels is calculated, and the determination result is output. This operation is performed while shifting the comparison start position (upper left end) pixel by pixel, and when a matching rate of a certain value or more is obtained, the same image as the reference image is displayed in the two-dimensional image at the shift position at that time. Suppose it exists.

[0009]

In the conventional pattern matching processing, since the image signal of the two-dimensional image is temporarily stored in the frame memory, a large capacity memory is required, and the reference image and the two-dimensional image are stored. Since the degree of overlap between the two is checked by relatively moving them one pixel at a time, it takes a lot of time for the pattern matching process after the two-dimensional image is obtained, and the time interval of the two-dimensional image used for the pattern matching process is large. I have no choice.

Further, if the pattern matching process including the color components is performed, it is usually necessary to perform the process for at least three colors, so that the pattern matching process takes further time.

For example, in the example of FIG. 11, as shown in FIG. 11C, when the reference image is vertically shifted by 64 lines, that is, when 512 × 62 pixels are shifted, a matching rate of a certain value or more. Is obtained, the processing time of the MPU per instruction step is 0.1 μS (for example,
Hitachi: In case of H8 / 300H series), the processing time at each pixel position is (0.1 x number of command steps at 1 pixel position) μS, and the binary image signal is written to the frame memory 3. After that, the time T from when the two images match is T = 0.1 × 512 × 64 × (the number of instruction steps at one pixel position) ≈3.3 × (the number of instruction steps at one pixel position) ) MS. If the two images are matched when they are shifted to the bottom of the two-dimensional image, T =
0.1 × 512 × (512-128) × (the number of command steps at one pixel position) ≈20 × (the number of command steps at one pixel position) mS, and until a match is obtained in the case of a color image Approximately 60 x (number of instruction steps at 1 pixel position)
It takes as long as ms.

As shown in the above example, the conventional pattern matching process requires a lot of time, and the pattern matching process may not be completed during the vertical blanking period or even during one vertical scanning period (about 33 mS). Considering the application of image recognition by pattern matching processing to crime prevention equipment, disaster prevention equipment, robots, etc., the two-dimensional image provided for processing is close to real time obtained at the smallest time interval. desirable.

An object of the present invention is to determine the difference between an image pattern in a two-dimensional image and a reference image at a high speed without using a frame memory, obtain a small time interval, and obtain a near real time image. An object of the present invention is to provide a pattern matching processing method capable of determining the difference between both patterns based on a three-dimensional image.

[0014]

In order to solve the above problems, the present invention binarizes an image signal obtained by raster-scanning a two-dimensional image to obtain a binarized image of an image pattern in the two-dimensional image. The start position and stop position on each raster scanning line in the signal are detected, the detected start position and stop position are standardized to obtain the standardized start position and stop position, and the standardized start position and stop position for the image pattern are obtained. The first feature is that the standardized start position and the stop position of the reference image pattern stored in advance are compared, and the difference between the image pattern and the reference image pattern is determined based on the comparison result.

A second feature of the present invention is that the standardization is such that the start position on the uppermost line where the pattern is raster-scanned is zero. Further, the present invention has a third feature in that the stop position is represented by a distance from the start position. Further, the present invention has a fourth feature in that the distance is represented by the number of pixels. Further, the present invention has a fifth feature in that the two-dimensional image is a color image and the pattern matching process is performed for each color component of the color image.

According to the first feature, the difference between the image pattern in the two-dimensional image and the reference image pattern can be determined at high speed without using a frame memory, and the pattern matching device and the two-dimensional image can be detected. Even when the target area is moving relatively, it is possible to compare / determine the difference between the two patterns based on the two-dimensional image obtained at a small time interval and close to real time. Further, according to the second to fourth characteristics, it is possible to easily compare the image pattern in the two-dimensional image with the reference image pattern by using a counter or the like. Furthermore, according to the fifth feature,
Since the determination is performed by including the color components, the accuracy can be improved.

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a pattern matching processing device to which the present invention is applied. The same reference numerals as those in FIG. 9 indicate the same or the same things.

In FIG. 1, a certain area is imaged by the image pickup section 1, and a grayscale image signal of a two-dimensional image obtained by this is input to the image processing section 2. The image processing unit 2 subjects the image signal to processing such as image enhancement and binarization, and the output binarized image signal is input to the start / stop position detection unit 9. Start / stop position detector 9
Detects a start position and a stop position on each raster scan line of each image pattern in the two-dimensional image based on the input binarized image signal.

The detected start position and stop position are input to the comparison / determination unit 10. Comparison / determination unit 1
0 standardizes the input start position and stop position to obtain the standardized start position and stop position, and the standardized start position and stop position for each image pattern and the standardized start position for the previously stored reference image pattern. The stop position is compared to determine whether each image pattern in the two-dimensional image is different from the reference image pattern.

FIG. 2 shows a specific example of the image processing unit 2 and the start / stop position detecting unit 9. For example, a grayscale image signal VD of 8 bits per pixel of a two-dimensional image from the image pickup unit 1 is an image. The comparator 6 of the processing unit 2 compares it with a threshold TH which is arbitrarily set, and becomes a binary image signal of 1 bit per pixel. Here, it is assumed that the binarized image signal has a white level in the image pattern. Incidentally, in the case of an image signal having a black level in the image pattern, a black-and-white inversion process may be performed, and the configuration may be adapted accordingly.

The binarized image signal is input in parallel to the white edge detection circuit 12 and the storage area switching circuit 13.
The horizontal synchronizing signal HSYN is input in parallel to the storage area switching circuit 13, the pixel counter 14 and the write address counter 17.

The V blanking period detection circuit 15 receives the horizontal synchronizing signal HSYN and the vertical synchronizing signal VSYN as input, detects the vertical blanking period, and outputs the detection signal to the comparison / determination unit 10. The vertical synchronization signal VSYN is also given to the storage area switching circuit 13.

The white edge detection circuit 12 detects the rising edge and the falling edge of the white level of the binarized image signal, and the count value of the pixel counter 14 at the time of detection is stored in the storage area 16- of the memory 16. 1, 16-
It is made to be written in any one of 2, 16-3, .... Each storage area 16-1, 16-2, 16-3 of the memory 16 ...
The storage order numbers 01, 02, 03, ... Denoting the delimiters of the storage areas are written in the leading addresses between. The writing position in each of the storage areas 16-1, 16-2, 16-3, ... Is the horizontal synchronization signal HSYN and the white edge detection circuit 12
Is specified by the write address counter 17 that receives the output of

The pixel counter 14 has a value of 1 from the image pickup unit 1.
It is cleared by the line image transfer end signal, that is, the horizontal synchronizing signal HSYN, and counts the image transfer clock (pixel clock) PCLK, and the count value represents the position on each line of the raster scanning.

When there are a plurality of image patterns in one screen, the storage area switching circuit 13 stores the storage areas 16-1 and 16 corresponding to the start position and the stop position on each raster scanning line for each image pattern. -2, 16-
For example, by paying attention to the continuity of figures, the orthogonal projection in the horizontal direction and the vertical direction, that is, a histogram is obtained by a counter or the like to separate the existence regions of each pattern. The publicly-known technique of is used, and the storage area to be stored is switched accordingly. Note that the existence area of each image pattern must be recognized before the white edge detection circuit 12 detects the white edge, but the existence area of the image pattern and the white edge may be recognized in successive frames. .

Further, in the case where an image pickup object has only one image pattern in the horizontal direction, a break in the pattern in the vertical direction is detected, and the image pattern in the horizontal direction is When there are a plurality of patterns without protruding from the pattern on the left side, the storage area switching circuit 13 is configured by a counter that is cleared by the horizontal synchronization signal HSYN and that counts the rising edge of the image pattern and a decoder that decodes the count value of the counter. However, it is possible to detect the white edge while switching the storage areas in a single frame by using the output of the decoder as a signal for switching the storage areas.

Next, the image pattern "A" in FIG.
The operation of FIG. 2 will be described on the assumption that a two-dimensional image including “B” and “C” is given. Note that FIG. 3 also shows the orthogonal projection (histogram) in the horizontal direction and the vertical direction. The image signal obtained by raster-scanning the two-dimensional image is compared with the threshold value TH by the comparator 6 and becomes a binarized image signal. The white edge detection circuit 12 detects the rising edge and the falling edge of the binarized image signal. At the time of this detection, the pixel counter 14 has a count value indicating the pixel position of the edge portion on the raster scanning line.

Therefore, if the count value of the pixel counter 14 is written in the storage area 16-1 of the memory 16 by the detection output of the white edge of the image pattern "A", the storage area 1
A start position and a stop position on each raster scanning line of the image pattern "A" are written in 6-1. For the image patterns “B” and “C”, the storage area switching circuit 13 stores the storage areas 16-2 and 16-.
3 and the start position and stop position on each line of the raster scan are similarly written individually.

FIG. 4 shows storage areas 16-1, 16-2, 1
6-3, image patterns “A”, “B”,
Specific numerical examples of the start position (start PIX) and stop position (stop PIX) of "C" are shown. Line NO.
Is a serial number with the first line in each image pattern as 1. It should be noted that the size of each storage area is required for the time during which the white level of the binarized image signal is input, but for the purpose of reducing the capacity of each storage area, line thinning writing (for example, every 5 lines) or the first line is performed. Therefore, a method such as writing a fixed number of lines is also possible.

Next, the processing in the comparison / determination unit 10 will be described. The comparison / determination unit 10 is composed of, for example, an MPU that executes a process by software, and when a vertical blanking period detection signal is given, the input start position and stop position are standardized and standardized start position (start ADR ) And stop position (stop ADR), and compares the standardized start position and stop position for each image pattern in the two-dimensional image with the standardized start position and stop position for the reference image pattern stored in the reference memory in advance. Then, the difference between each image pattern in the two-dimensional image and the reference image pattern is determined.

First, the standardization processing of the start position and the stop position will be described. This standardization process is a process for facilitating comparison between each image pattern and the reference image pattern. For example, when the start position of the uppermost line of the image pattern is set to 0 This is a process in which the relative value is obtained, the stop position of each line is obtained as a distance such as the number of pixels from the start position of the corresponding line, and the relative value and the distance are written in the result memory 18.

FIG. 5 is a flowchart showing an example of a processing procedure for obtaining the standardized start position and stop position, and this procedure comprises the following steps S1 to S11.

(S1) The data in the storage area 01 of the memory 16 is read. (S2) It is determined whether the read data is the first line. (S3) If it is the first line, (stop position-start position) is calculated to obtain the standardized stop position (stop ADR) of the line. (S4) Standardized start position (start ADR) at the first line address of the first storage area 01 of the result memory 18
0 and the standardized stop position (stop ADR) obtained in step S3 are written. (S5) If the determination result in step S2 is not the first line, the start position (= start position of the first line−start position of the line) is calculated to obtain the standardized start position (start ADR) of the line. . (S6) Subsequently, a stop position (stop position of the line-start position of the line) is calculated to obtain a standardized stop position (stop ADR) of the line. (S7) The standardized start position and standardized stop position obtained in steps S5 and S6 are written to the next line address in the first storage area 01 of the result memory 18. (S8) It is determined whether or not all the lines in the first storage area have been completed. (S9) If it is determined in step S8 that all the lines in the first storage area of the memory 16 have not been completed, the read address is updated, and the processes from step 1 onward are repeated for the next line. (S10) If it is determined in step S8 that all the lines in the first storage area of the memory 16 are finished, it is determined whether or not the standardization of the data in all the storage areas of the memory 16 is finished, and the processing is finished. If it is determined that the standardization process is completed. (S11) If it is determined in step S10 that the standardization has not been completed, the standardization process in step S2 and subsequent steps is repeated for the data in the next storage area of the memory 16.

FIG. 6A shows the start position (start ADR) and stop position (stop ADR) for the image patterns "A", "B", and "C" written in the result memory 18 after the standardization process. It becomes like (b) and (c). In the above example, the stop position of each line is represented by the number of pixels from the start position of the line, but this is represented by a relative value when the start position of the first line of the image pattern is 0. You may do it.

Next, the standardized start position and stop position for each image pattern are compared with the start position and stop position of the reference image pattern stored in advance, and each image pattern in the two-dimensional image and the reference image pattern are compared. The process of determining the difference will be described with reference to FIG. 7. The figure also shows the start and stop positions of the image patterns "A", "B", and "C" before standardization.

The reference memory 19 stores the standardized start position and stop position of the reference image pattern as shown in FIG. 6D, for example, as with each image pattern in the two-dimensional image. It can also be obtained by capturing a reference image prior to the pattern matching process. Here, the reference image pattern is the image pattern “A”.
Specific values are given assuming that they are the same as.

The result memory 18 stores the standardized start position and stop position of each image pattern obtained by the above standardization process, and the head address between the respective storage areas for each image pattern is the same as that of the memory 16. To
Storage order numbers 01, 02, 03, which indicate the division of the storage area,
・ ・ Is entered.

FIG. 7 is a functional block diagram showing an example of the comparison / determination unit 10, which is composed of an MPU that executes processing by software. FIG. 8 is a flowchart showing an example of a comparison / determination processing procedure, and this procedure includes the following steps S101 to S110.

(S101) First, reading of the start and stop positions of the first line of the area 01 of the result memory 18 and the start and stop positions of the first line of the reference memory 19 is started. (S102) The comparing unit 20 compares the read start positions and the read stop positions. (S103) If the comparison results match, the count value of the match counter 21 is incremented by one. (S104) The counter value of the counting counter 22 that counts the number of comparison lines is incremented by one. (S105) It is determined whether or not the data of the next line address in the result memory 18 is the storage sequence number 01, 02, ... This is a judgment as to whether or not all the data in the storage area 01 has been read. (S106) If the determination result of step S105 is not the storage order number, the result memory 18 and the reference memory 19
Line address is updated, and the processes in and after step 102 are repeated. (S107) If the determination result in step S105 is the storage order number, the comparison of the data in the storage area 01 is completed, and the count value of the match counter 21 and the count value of the count counter 22 are sent to the match rate calculation unit 23. The data is read and the ratio between them (count value of coincidence counter / count value of counter) is calculated. (S108) It is determined whether the ratio calculated in step S107 is 90% or more, for example. It should be noted that 90% is only an example, and if this value can be variably set externally, it is possible to make a determination according to the influence of noise or the like and the required accuracy. (S109) When it is determined in step S108 that 90% or more, the image pattern of the result memory 18 and the reference memory 1
The notification of the coincidence with the reference image pattern of No. 9 is notified, and the process ends. (S110) If it is not determined to be 90% or more in step S108, the processes in step S101 and subsequent steps are repeated for the data in the next storage area of the result memory 18.

The judgment result of the comparison / judgment processing can be notified by, for example, a one-frame image transfer end signal given from the image pickup unit 1 after the comparison / judgment processing, that is, a vertical synchronization signal VSYN.

Next, the processing time in the present invention will be described. Here, the processing time of the MPU per one step of the instruction is set to 0.1 μS as in the prior art, and as shown in FIG.
As shown in FIG. 11 and FIG. 11, it is assumed that there are three image patterns in the two-dimensional image and data of 10 lines is stored for each image pattern. The number of lines of data to be stored per image pattern can be appropriately thinned out, and if the data of the reference image pattern to be compared is also thinned out, there is no problem in comparison.

Generally, the processing time T is represented by T = (processing time of MPU per instruction step × number of steps of one processing × total number of processing). The process of obtaining the standardized start position and stop position and writing them in the result memory 18 is a maximum of 7 steps per line (S2-S5-S).
6-S7-S8-S10-S11) and processing of 10 lines is required for each of the three image patterns, the processing time T1 required for standardization is T1 = 0.1.
(ΜS) × 7 × 10 × 3 = 0.021 (mS).

The steps of the comparison process in the comparison / judgment process are 6 steps per line (S101 to S101).
106), and the steps of the determination process are three steps (S107-S108-S109 or S107-S).
108-110), the processing time T2 required for comparison / determination is T2 = 0.1 (μS) × 6 × 10 × 3 +
It becomes 0.1 (μS) × 3 × 3 = 0.189 (mS).

The total processing time T is T = T1 + T2 =
0.0399 (mS), which is less than one horizontal line period (0.064 (mS)). Therefore, according to the present invention, the processing can be performed in the vertical blanking period having a period of one horizontal line or more, and the pattern matching process can be executed based on the subsequently input two-dimensional image.

Next, the application of the present invention to alarm notification in a vehicle will be briefly described. The device to which the present invention is applied is mounted on a vehicle with the imaging unit 1 oriented so as to output an image signal of a two-dimensional image of a region including a traffic signal. The image pickup unit 1 outputs a color image signal.
Also, the device may be portable rather than fixedly mounted in the vehicle. The reference image pattern is, for example, a red signal pattern, and the start position and the stop position of each raster scanning line are stored in the reference memory 19.

The start position and the stop position in each raster scanning line for each color component of the color image signal output from the image pickup unit 1 are detected and standardized, and the standardized start position and stop for the reference image pattern, that is, the red signal pattern are detected. The position is compared to determine whether they match. Since the signal magnitude, pattern, and its color of the traffic light are known in advance, if the vehicle runs at a predetermined distance from the traffic light and the shape, visual magnitude, and color of the signal pattern match the reference pattern, both Matching of the start position and the stop position on each line of the raster scan of the pattern is determined, and the fact is notified. That is, when the vehicle approaches the red light, the fact can be notified when the vehicle approaches the predetermined distance.

The distance from the traffic light to be notified can be changed by changing the size of the reference image pattern. Note that the pattern of the yellow signal may be determined, and the notification content may be different for the blue signal, the yellow signal, and the red signal by determining the color based on the image signal. Further, when the image pickup is affected by ambient light or the like, well-known light amount compensation means such as dynamic bias control may be adopted.

In the present invention, not only the vehicle as described above, but also a worker in a plant, an old man or a child, or a physically handicapped person can detect his or her own danger early by appropriately selecting a reference pattern. It can be applied to crime prevention / disaster prevention equipment and robots to ensure safety.

The device to which the present invention is applied may not be exclusively configured for a vehicle, a plant, or an individual, but may be configured to be portable so as to have a processing program, a reference pattern, and a notification. You can rewrite the contents,
If it can be changed by replacing the ROM, etc., it can be configured as a general-purpose device that can be appropriately used in a necessary place.

[0050]

As is apparent from the above description, according to the present invention, it is possible to compare and determine the difference between the image pattern in the two-dimensional image and the reference image pattern without using a frame memory. Since the device can be configured with portable small hardware and the comparison and determination can be performed at high speed, the pattern matching device and the target area of the two-dimensional image are relatively moved. Also in this case, it is possible to compare and determine the difference between both patterns based on a two-dimensional image obtained at small time intervals and close to real time.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a pattern matching processing device to which the present invention is applied.

FIG. 2 is a block diagram showing a specific example of an image processing unit and a start position / stop position detection unit in FIG.

FIG. 3 is an explanatory diagram of an example of a two-dimensional image.

4 is an explanatory diagram of an example of specific numerical values of a start position and a stop position of an image pattern in the two-dimensional image of FIG.

FIG. 5 is a flowchart showing an example of a processing procedure for obtaining a standardized start position and a stop position.

FIG. 6 is an explanatory diagram of an example of standardized start positions and stop positions for an image pattern and a standard image pattern in a two-dimensional image.

FIG. 7 is a functional configuration diagram showing an example of a comparison / determination unit in FIG. 1.

FIG. 8 is a flowchart illustrating an example of a comparison / determination processing procedure.

FIG. 9 is a block diagram of an apparatus to which a conventional pattern matching processing method is applied.

FIG. 10 is a detailed block diagram of the frame memory of FIG. 9 and its peripheral circuits.

FIG. 11 is an explanatory diagram of a comparison / determination process in FIG.

[Explanation of symbols]

1 ... Imaging unit, 2 ... Image processing unit, 3 ... Frame memory, 4 ... Reference memory, 5, 10 ... Comparison / determination unit, 6,
20 ... Comparator, 7 ... Write timing control circuit, 8
... MPU, 9 ... Start / stop position detector, 12
... White edge detection circuit, 13 ... Storage area switching circuit, 14
... Pixel counter, 15 ... V blanking period detection circuit, 16 ... Memory, 17 ... Write address counter, 18 ... Result memory, 19 ... Reference memory, 21 ...
Matching counter, 22 ... Counting counter, 23 ... Matching rate calculation unit

Claims (5)

[Claims] 1. A pattern matching processing method for determining the difference between an image pattern in a two-dimensional image and a reference image pattern, wherein an image signal obtained by raster-scanning a two-dimensional image is binarized, A start position and a stop position on each raster scanning line in the binarized image signal of the image pattern are detected, and the detected start position and stop position are standardized to obtain a standardized start position and a stop position. Pattern matching characterized in that the standardized start position and stop position are compared with the standardized start position and stop position for a reference image pattern stored in advance, and the difference between the image pattern and the reference image pattern is determined by the comparison result. Processing method. 2. The pattern matching processing method according to claim 1, wherein the standardization is to set the start position on the uppermost line where the pattern is raster-scanned to zero. 3. The pattern matching processing method according to claim 1, wherein the stop position is represented by a distance from the start position. 4. The pattern matching processing method according to claim 4, wherein the distance is represented by the number of pixels. 5. The pattern matching processing method according to claim 1, wherein the two-dimensional image is a color image, and pattern matching processing is performed for each color component of the color image.