JP2000182056A - Picture processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To highly precisely maintain positioning precision and to realize high speed pattern matching with software by reconstituting a picture formed only of a picture area which largely affects the calculated result of a correlation operation and executing the correlation operation on the reconstituted reconstitution picture. SOLUTION: A picture area which largely affects the calculated result of a correlation operation is the part 400 of the outline of a circle picture 400A and the outline part is considered to be an area where a luminance difference with a peripheral picture is large, for example. A differential filter is applied to a reference picture and a differential picture 401 is obtained. The differential picture is divided into the small areas 402-417 of appropriate forms and the average values or the maximum values of the differential value are obtained for the respective small areas. The luminance difference in the small areas is judged with the obtained average values or the maximum values of the respective small areas. Highest n-pieces of small areas among the obtained plural average values or maximum values are made into partial pictures for reconstituting the picture as the areas (small areas 408, 411, 413, 416 or the like) where the luminance difference with the peripheral pictures is large.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly, to an image processing apparatus capable of performing pattern matching for comparing a reference image and a target image by a correlation operation.

[0002]

2. Description of the Related Art In recent years, in the industrial field, automation devices using image processing technology for measurement and inspection have been widely used.
In particular, an image processing method called pattern matching is a basic image processing method among automation devices using image processing technology. The conventional pattern matching method is performed in the following manner. That is, first, an image as clear as possible is photographed for an object to be measured and inspected. Next, a reference image including a portion indicating the characteristics of the target object is extracted from the captured image of the target object. Next, the whole image of the inspection object to be actually inspected is photographed by a camera, and an image to be compared is extracted from this image as a comparison image. From the reference image and the comparison image, a correlation operation is performed to calculate a coincidence rate and the like. Further, by repeatedly performing the above-described comparison image extraction processing while changing the comparison position, the position of the region including the comparison image that best matches the reference image in the inspection target image is searched, and the obtained position is used. Then, the displacement is corrected. In general, a correlation operation is used for comparison between the reference image and the comparison image.

However, in this operation, it is necessary to perform an operation relating to luminance and the like for each pixel corresponding to the reference image and the comparison image, and further, a process of obtaining a matching rate by a correlation operation is repeatedly performed for a continuous area. Therefore, there is a problem that a very large number of calculations are repeated and the processing time becomes long. In order to solve this problem, Japanese Patent Application Laid-Open No. 8-189820 (hereinafter referred to as a first prior art) and Japanese Patent Application Laid-Open
No. 22090 (hereinafter referred to as second prior art)
It has been known. In the image processing apparatus according to the first related art, by removing redundant image areas that hardly affect the calculation result of the correlation calculation, the calculation is speeded up while maintaining the positioning accuracy of the pattern matching. I can do it. In the second prior art,
A local feature extraction process for converting a target image to be subjected to pattern matching into feature element arrangement information, which is data obtained by combining a smaller number of preset feature elements and rough spatial arrangement information of the feature elements, is performed. Will be By performing pattern matching using the above-described feature element arrangement information having a smaller amount of information than the target image, the calculation cost can be reduced, and image recognition by pattern matching can be performed on a deformed or enlarged / reduced figure.

[0004]

In the first prior art, in order to remove a redundant image area, it is necessary to set a free area other than a rectangle when setting a reference image. For that purpose, it is necessary to introduce a new relative coordinate memory. Further, for the captured original image, it is necessary to determine to which position in the array address the luminance data of each pixel is to be substituted for each pixel by referring to the relative coordinate memory. Therefore, in the first technique, the calculation cost of the correlation operation itself is reduced, but instead, the amount of calculation processing of referencing the relative coordinate memory and substituting the luminance data into the array address becomes enormous. Issues arise. This makes it difficult to configure the device by software for cost reduction. In addition, the setting of the characteristic portion of the reference image using a free shape other than a rectangle is performed artificially, and there is a problem that the setting largely depends on the knowledge and experience of the human being to set.

In the second prior art, since the characteristic element arrangement information is used, the calculation cost at the time of the correlation operation is reduced. However, since an algorithm for performing the above-described local feature extraction processing is newly added, speeding up of the entire pattern matching processing is not always realized. In general, there are various features in the target image of the local feature extraction processing, and when the feature elements are changed in accordance with the target image, as in the first related art, the feature elements are changed. When setting the arrangement information, there is a problem that the setting largely depends on the knowledge and experience of the person who sets the characteristic element arrangement information. Further, when performing the local feature extraction processing, it is necessary to divide an image into small areas and process the data in order to decompose the data into the feature element units. Therefore, only the spatial resolution in the unit of the small area can be secured with the feature element arrangement information. Therefore, even if pattern matching is performed precisely and precisely, there is a problem that positioning accuracy can be guaranteed only up to the unit of the size of the small area. According to the present invention, there is no need to introduce dedicated hardware for speeding up, no need to introduce complicated calculation processing such as an image recognition algorithm for feature extraction for each target image, and knowledge and experience of image processing. Provided is an image processing apparatus capable of performing a process of extracting a partial image of a reference image regardless of image quality, reducing calculation cost while maintaining high positioning accuracy, and performing high-speed pattern matching. It is.

[0006]

According to a first aspect of the present invention, there is provided an image processing apparatus for comparing a reference image stored in advance with a target image by a correlation operation. Partial image extracting means for extracting a plurality of predetermined partial images constituting the reference image, partial image reconstructing means for reconstructing the extracted partial images into one reconstructed image and storing the same, and extracting the partial images Procedure storing means for storing the procedure from step to reconstruction, target image reconstructing means for reconstructing and storing a target image according to the procedure stored in the procedure storing means, and stored in the partial image reconstructing means. An image processing apparatus comprising: a correlation operation unit that performs a correlation operation between the reconstructed image of the reference image and the reconstructed image of the target image stored in the target image reconstruction unit. It is constructed by.

According to a second aspect of the present invention, in the image processing apparatus according to the first aspect, the partial image extracting means includes:
The gist is that an image is extracted using the luminance difference of the image as an index. According to a third aspect of the present invention, in the image processing apparatus according to the first aspect, the partial image extracting means creates a differential image relating to the brightness of the image, and divides the differential image into a plurality of appropriately shaped small regions. The gist is that the plurality of partial images are extracted by calculating the average or the maximum value of the differential values for each of the small areas. According to a fourth aspect of the present invention, in the image processing apparatus according to the first aspect, the partial image extracting unit performs a predetermined process on the reference image to obtain a center of a characteristic image included in the reference image. The gist is to obtain coordinates and extract the plurality of partial images including the edge of the feature image existing around the center coordinates according to a predetermined rule. Further, the invention according to claim 5 is based on claim 4.
In the image processing apparatus described above, the gist is that the predetermined process is a process of extracting the characteristic image that satisfies a preset condition. According to a sixth aspect of the present invention, in the image processing apparatus according to the first aspect, the partial image extracting means controls a pointer capable of designating an area on the image by moving on the image. It is the gist of the input / output device. According to a seventh aspect of the present invention, in the image processing apparatus according to any one of the first, second, third, fourth, fifth and sixth aspects, the target image reconstructing means reconstructs the target image. At this time, the target image is scanned, and a plurality of reconstructed images of the target image are sequentially generated and stored, and the correlation operation means reconstructs and stores the reconstructed image stored by the partial image reconstructing means. The gist is that correlation calculation is performed on the constituent image and the plurality of reconstructed images.

The invention according to claim 8 is the image processing apparatus according to claim 7, wherein the correlation operation means includes:
Image recognition means for determining a coincidence rate between the reconstructed image of the reference image and the plurality of reconstructed images for the target image, and recognizing the same image as the reference image in the target image from the obtained plurality of coincidence rates. The gist is to have it. The invention according to claim 9 is based on claims 1, 2, 3,
In the image processing apparatus according to any one of 4, 5, 6, 7, and 8, a constant part used only for the correlation operation and determined only in advance by the reference image is calculated and stored when the reference image is reconstructed. The gist is that a constant part calculating means is further provided. Claims 1, 2, 3, 4, 5, 6,
According to the image processing apparatuses described in 7, 8, and 9, an image consisting only of image regions that greatly affect the calculation result of the correlation operation is reconstructed, and the correlation operation is performed on the reconstructed reconstructed image. As a result, the calculation of the redundant image area can be omitted, and the calculation cost related to the correlation calculation can be reduced. In addition, since the partial image is reconstructed into one reconstructed image, dedicated hardware such as a relative coordinate memory is not required, and the previously stored image is partially retained as it is. High positioning accuracy can be maintained. As a result, high-speed pattern matching can be realized by software. In addition, according to the image processing apparatus of the second, third, fourth and fifth aspects, when extracting a plurality of predetermined partial images constituting the reference image from the reference image, it is automatically performed according to a predetermined procedure. Extraction eliminates the need for a human to determine which image portion greatly affects the correlation calculation as in the conventional case. Therefore, even a person who does not have special knowledge can extract an image portion that greatly affects the correlation calculation. Further, according to the image processing apparatus of the present invention, when extracting a plurality of predetermined partial images constituting the reference image from the reference image, the partial images are determined according to the idea of the person who performs the work. be able to. Therefore, it is possible to more accurately extract a partial image that a person who has a great deal of knowledge and experience in image processing considers to greatly affect the correlation calculation. Claim 7,
According to the image processing apparatus described in Item 8, it is possible to recognize the reference image in the target image, and thus it is possible to determine the existence, position, and deviation of the reference image in the target image. In addition, according to the image processing apparatus of the ninth aspect, the same calculation is repeatedly performed because the constant portion used for the correlation operation and determined only in advance by the reference image is calculated and stored when the reference image is reconstructed. No need,
Calculation cost can be reduced.

[0009]

Embodiments of the present invention will be described below with reference to the accompanying drawings to provide an understanding of the present invention.
It should be noted that the following embodiments are examples embodying the present invention, and do not limit the technical scope of the present invention. Here, FIG. 1 is a functional block diagram schematically showing an image processing apparatus according to one embodiment of the present invention, FIG. 2 is a hardware block diagram schematically showing an image processing apparatus according to one embodiment of the present invention, and FIG. Is a flowchart showing a job flow when setting a reference image, FIG. 4 is a conceptual diagram showing a method for extracting a first partial image, FIG. 5 is a conceptual diagram showing a method for extracting a second partial image, and FIG. FIG. FIG. 7 is a flowchart showing a job flow of pattern matching, and FIG. 8 is a flowchart showing a job flow of data determination processing in pattern matching.

As shown in FIG. 1, an image processing apparatus according to the present embodiment stores a measurement / inspection object 1, a camera 2 for photographing the measurement / inspection object 1, and an image photographed by the camera. This is the same as the related art in that an image storage unit 3 for performing the comparison and a correlation operation unit 8 for comparing the stored images are held. The image processing apparatus according to the present embodiment is different from the conventional one in that the image processing apparatus further includes a partial image extracting unit 4, a procedure storing unit 5, a partial image reconstructing unit 6, and a target image reconstructing unit 7. From the reference image stored in the storage unit 3,
The partial image extracting section 4 extracts a plurality of predetermined partial images constituting the reference image, and the partial image reconstructing section 6 reconstructs the extracted partial images into one reconstructed image. The procedure from the extraction to the reconstruction of the partial image is stored by the procedure storage unit 5, and the procedure storage unit 5 is stored by the target image reconstruction unit 7.
In that the target image stored in the image storage unit 3 is reconstructed and stored in accordance with the procedure stored in the image storage unit 3, and that the correlation operation unit 8 stores the reference image stored in the partial image reconstruction unit 6. The point is that a correlation operation is performed between the reconstructed image and the reconstructed image of the target image stored in the target image reconstructing unit 7.

Next, details of the image processing apparatus will be described. Further, in order to contribute to understanding, the correlation calculation is, for example, a normalized correlation calculation for obtaining a coincidence rate α defined by the following equation.

(Equation 1) The image processing apparatus according to the present embodiment is embodied by, for example, hardware as shown in the block diagram of FIG. First, the configuration of the image processing apparatus according to the present embodiment will be described with reference to FIGS. The image processing device is
It comprises a CPU 201, a ROM 202, a storage unit 203, interfaces 204, 205 and 206, and a bus 209.
Connected by Each of the above interfaces 20
The camera 2, input device 207, and output device 208 are connected to 4, 205, and 206, respectively. Further, the storage unit 203 stores the image storage unit 3 and the procedure storage unit 5.
Is made up of Further, the partial image extraction unit 4,
The program for achieving the functions of the partial image reconstructing unit 6, the target image reconstructing unit 7, and the correlation calculating unit 8 is the RO program.
The program is stored in M202, and is read from the ROM 202 to the CPU 201 and executed according to the processing described below. Further, the image information including the position information associated with the storage processing in the partial image extraction unit 4, the partial image reconstruction unit 6, the target image reconstruction unit 7, and the correlation operation unit 8 is stored in the storage unit 203. .

Hereinafter, description will be made focusing on the processing by the CPU 201 shown in FIG. First, an outline of a procedure for setting a reference image in the present embodiment will be described with reference to FIGS. Here, S301 to S307 indicate processing procedure numbers. First, the measurement / inspection object 1 to be a target of the reference image is photographed by the camera 2 to obtain the reference image in the image storage unit 3 (S301). The partial image extracting unit 4 extracts a plurality of partial images constituting the reference image by a predetermined method described later (S302), and stores partial image information such as the position and size in the procedure storage unit 5 (S303). . Next, the partial image reconstruction unit 6 reconstructs and stores the plurality of partial images into one image using the partial image information stored in the procedure storage unit 5 (S304). The procedure from the extraction of the plurality of partial images to the reconstruction is stored in the procedure storage unit 5 (S305). In addition, the parameters N, T1, and T2 that depend only on the reconstructed image of the reference image in Equation 1 that is an equation for calculating the coincidence rate α are also calculated (S306) and held (S307). By the above operation, the reconstructed image of the reference image used for the correlation operation and the constant part used for obtaining the coincidence rate α are retained, and the procedure for reconstructing the target image is also stored.

Next, a method of extracting a plurality of predetermined partial images constituting the reference image from the reference image in the partial image extracting unit 4 will be described in detail with reference to three examples. First, a first method for extracting the partial image will be described with reference to FIG. As a means for extracting the plurality of partial images, a method using luminance can be considered. An image area that greatly affects the calculation result of the correlation operation is, for example, a contour portion 400 of the circle image 400A. This contour portion is considered to be a region having a large luminance difference from the surrounding image. A differential image 401 is obtained by performing a differential filter. Next, the differential image is divided into small regions 402 to 417 having an appropriate shape, and an average value or a maximum value of the differential values is obtained for each of the small regions. The luminance difference in the small area can be determined from the average value or the maximum value of the obtained small areas, and it can be said that the larger the value, the larger the luminance difference in the small area. Next, among the plurality of average values or the maximum values obtained, the top n small regions are determined to be regions having a large luminance difference from the surrounding image (in FIG. 4, the small regions 408, 41).
1, 413, 416, etc.) as partial images for image reconstruction.

Next, a second method for extracting the partial image will be described with reference to FIG. As a means for extracting the plurality of partial images, a method using the characteristic of the shape of the characteristic portion of the reference image can be considered. This method is advantageous when the feature portion of the reference image is a simple feature image such as a circle or an ellipse. First, a binarized image is obtained from the reference image using an appropriate binarization threshold. In order to discriminate the feature image from the binarized images, labeling is performed by attaching a sign such as a number to each binarized image, and label images 501 and 50 are obtained.
2,503 are obtained. Further, in the label image, a feature amount such as an area and a radius of curvature of each label figure is obtained, and a label figure 501 that satisfies a preset condition of the feature image
The label number 1 of (in this case, a characteristic image of a circle) is obtained, and this label image 501 is set as a characteristic image. Thereby, the shape information of the characteristic image in the reference image can be obtained. Next, in order to determine end coordinates, which are boundaries between the feature image and the other images in the reference image, barycentric coordinates 509 of the figure whose label number has been obtained are obtained. The end coordinates 511, 512, 513, and 514 of the label graphic are scanned in four directions. In scanning, the edge coordinates can be determined by monitoring the label number of the pixel being scanned. In this way, the end coordinates of the four points are obtained, and the rectangular small areas 505, 506, 507, and 508 centered on the end coordinates are set as partial images for image reconstruction. Here, the preset feature image may be a rectangle or a triangle other than a circle and an ellipse, and a partial image is extracted by a method suitable for the shape of the feature image.

Next, a third method for extracting the partial image will be described. As means for extracting the plurality of partial images, an input / output device such as a mouse or a keyboard for controlling a pointer capable of designating an area on the image by moving on the image, and a monitor for displaying a reference image Can be used. In this case, for example, the operator
A rectangular area to be extracted is set by designating two points of a certain vertex and a diagonal vertex of the rectangular area with the input device. The specified rectangular portion is a partial image for image reconstruction. The area to be specified may not be a rectangle, but is desirably a simple figure after reconstruction. In the case of the first and second methods for extracting the partial image,
Even a person who does not have special knowledge can extract an image portion that greatly affects the correlation calculation. Furthermore, according to the third method of extracting the partial image, since a person with abundant knowledge and experience in image processing intervenes,
It is possible to extract a partial image that is considered to have a great influence on the correlation operation with good accuracy, and it is possible to reduce the load on program creation.

Next, referring to FIG. 6, a method for reconstructing the extracted partial image into one reconstructed image in the partial image reconstructing unit 6 will be described in the second section for extracting the partial image.
This will be described in detail by using the method described above. The partial image extracted by any or a combination of the first to third methods for extracting the partial image (the second method in this example) is stored in the procedure storage unit 5 as coordinate information in the reference image. Have been. The partial image reconstruction unit 6 converts the partial images 505 and 50 from the reference image using the coordinate information.
6, 507, 508 are obtained, the partial images 505, 506 are made to approach the center of the circle, reconstructed and stored as one rectangular image 601. In the case where the partial image is a square or the like, there is no problem even if the reconstructed image is a rectangle in which the squares are arranged in a horizontal line. Next, a detailed procedure of pattern matching for extracting the same image as the reference image in the target image will be described with reference to FIGS. Here, S701 to S707 and S801 to S803 indicate processing procedure numbers. First, as shown in FIGS. 1 and 7, a measurement / inspection object 1, which is an object image to be compared with a reference image, is photographed by a camera 2, and an object image is obtained in an image storage unit 3 (S
701). Next, in the target image reconstructing unit 7, in order to manage the scanning position on the target image, the extraction position of the image area on the target image to be processed by the partial image reconstructing unit 6 is stored (S702). ). Further, the reference image reconstruction procedure stored in the procedure storage unit 5 (S703)
Is applied to the target image, the target image is decomposed into partial images of the target image, and then reconstructed and stored (S704).
Next, in the correlation operation unit 8, the already calculated parameters N, held in the partial image reconstruction unit 6,
Using T1 and T2, a correlation operation is performed between the reconstructed image of the reference image and the reconstructed image of the target image stored in the target image reconstructing unit 7 to obtain a matching rate at that time (S70).
5). In this way, by reconstructing each image into one image in which partial images composed of simple components are combined, position information of countless pixel data to be referred to in the reference image and the target image at the time of correlation calculation is obtained. There is no need to consider it, and redundant portions can be removed, so that high-speed correlation calculation can be performed. Next, data determination processing (S70
6) is performed. This will be described in detail with reference to FIG. First, it is determined whether the coincidence rate α obtained by the correlation operation unit 8 is the maximum (S801). If the coincidence rate is the maximum, the coincidence rate α at that time and the extraction position stored in the scanning processing (S702) are stored and held (S802), and it is determined whether the scanning is completed (S803). If the coincidence rate is not the maximum, the end of scanning is determined as it is (S803).
In the determination of the scanning end, it is determined whether or not the extraction position has reached the final extraction position in the target image. If the scanning is not completed, the extraction position of the target image is changed in the scanning process (S702), and the process is repeated from the target image reconstruction process (S704). The obtained data is output to a connected display or the like by the determination data output process (S707).

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention may be implemented with various improvements, and the correlation operation between two images is not limited to the normalized correlation operation. Each means (partial image extraction unit 4, procedure storage unit 5, partial image reconstruction unit 6, target image reconstruction unit 7)
Is achieved by software, but the same function can be achieved by hardware.

[0018]

According to the first, second, third, fourth, fifth, and sixth aspects of the present invention.
According to the image processing apparatuses described in 7, 8, and 9, an image consisting only of image regions that greatly affect the calculation result of the correlation operation is reconstructed, and the correlation operation is performed on the reconstructed reconstructed image. As a result, the calculation of the redundant image area can be omitted, and the calculation cost related to the correlation calculation can be reduced. In addition, since the partial image is reconstructed into one reconstructed image, dedicated hardware such as a relative coordinate memory is not required, and high positioning accuracy can be maintained. As a result, high-speed pattern matching can be realized by software. Moreover, claims 2, 3, 4,
According to the image processing apparatus described in Item 5, when extracting a plurality of predetermined partial images constituting the reference image from the reference image, the partial images are automatically extracted according to a predetermined procedure. There is no need for a human to judge whether the part greatly affects the correlation calculation. Therefore, even a person who does not have special knowledge can extract an image portion that greatly affects the correlation calculation. Further, according to the image processing apparatus of the present invention, when extracting a plurality of predetermined partial images constituting the reference image from the reference image, the partial images can be determined according to the idea of a person who performs the work. . Therefore, humans with rich knowledge and experience in image processing can more accurately
It is possible to extract a partial image that is considered to greatly affect the correlation calculation. According to the image processing apparatus of the present invention, since the reference image can be recognized in the target image, it is possible to determine the existence, position, and deviation of the reference image in the target image. it can. In addition, according to the image processing apparatus of the ninth aspect, the same calculation is repeatedly performed because the constant portion used for the correlation operation and determined only in advance by the reference image is calculated and stored when the reference image is reconstructed. There is no need, and the calculation cost can be reduced.

[Brief description of the drawings]

FIG. 1 is a functional block diagram schematically showing an image processing apparatus according to an embodiment of the present invention.

FIG. 2 is a hardware block diagram schematically illustrating an image processing apparatus according to an embodiment of the present invention.

FIG. 3 is a flowchart illustrating a job flow when a reference image is set.

FIG. 4 is a conceptual diagram showing a method of extracting a first partial image.

FIG. 5 is a conceptual diagram showing a method of extracting a second partial image.

FIG. 6 is a diagram showing the concept of image reconstruction.

FIG. 7 is a flowchart illustrating a job flow of pattern matching.

FIG. 8 is a flowchart illustrating a job flow of data determination processing in pattern matching.

[Explanation of symbols]

4 ... partial image extracting unit (corresponding to partial image extracting means) 5 ... procedure storage unit (corresponding to procedure storing means) 6 ... partial image reconstructing unit (corresponding to partial image reconstructing means) 7. ..Target image reconstructing unit (corresponding to target image reconstructing means) 8... Correlation calculating unit (corresponding to correlation computing means)

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Katsuya Takaoka 1-5-5 Takatsukadai, Nishi-ku, Kobe City, Hyogo Prefecture Inside Kobe Research Institute, Kobe Steel Ltd. (72) Inventor Sadaharu Baba Kinuga, Moka City, Tochigi Prefecture No. 15 Kobe Steel Moka Works (72) Inventor Keiji Okada 15 Kinuigaoka, Moka City, Tochigi Pref. DC39 5C076 AA19 AA36 CA02 5L096 CA24 EA35 FA32 FA34 FA62 GA19 JA03 KA13 9A001 BB06 GG01 HH21 HH23

Claims (9)

[Claims] 1. An image processing apparatus for comparing a reference image stored in advance and a target image by a correlation operation, wherein partial image extracting means for extracting a plurality of predetermined partial images constituting the reference image from the reference image. A partial image reconstructing means for reconstructing the extracted partial image into one reconstructed image and storing the reconstructed image; a procedure storing means for storing a procedure from extraction of the partial image to reconstructing; Target image reconstructing means for reconstructing and storing a target image in accordance with the stored procedure; and the reconstructed image of the reference image stored in the partial image reconstructing means and the reconstructed image stored in the target image reconstructing means. An image processing apparatus comprising: a correlation operation unit that performs a correlation operation with a reconstructed image of a target image. 2. The image processing apparatus according to claim 1, wherein said partial image extracting means extracts an image using a luminance difference of the image as an index. 3. The partial image extracting means creates a differential image relating to the luminance of the image, divides the differential image into a plurality of small regions having an appropriate shape, and calculates an average or a maximum value of the differential values for each of the small regions. The image processing apparatus according to claim 1, wherein the plurality of partial images are extracted by calculating the following. 4. The partial image extracting means obtains center coordinates of a feature image included in the reference image obtained by performing a predetermined process on the reference image, and determines the center coordinates around the center coordinates according to a predetermined rule. The image processing apparatus according to claim 1, wherein the plurality of partial images including an end line of the characteristic image to be extracted are extracted. 5. The process according to claim 4, wherein the predetermined process is a process of extracting the characteristic image satisfying a preset condition.
The image processing apparatus according to any one of the preceding claims. 6. An image processing apparatus according to claim 1, wherein said partial image extracting means is an input / output device such as a mouse for controlling a pointer capable of designating an area on said image by moving on said image. . 7. The target image reconstructing means scans the target image when reconstructing the target image, and sequentially generates and stores a plurality of reconstructed images for the target image. The correlation calculating means performs a correlation calculation between the reconstructed image reconstructed and stored by the partial image reconstructing means and the plurality of reconstructed images.
The image processing apparatus according to any one of 2, 3, 4, 5, and 6. 8. The correlation calculating means calculates a coincidence rate between the reconstructed image of the reference image and a plurality of reconstructed images of the target image, and obtains a reference image and a reference image in the target image from the obtained plurality of coincidence rates. The image processing apparatus according to claim 7, further comprising an image recognition unit that recognizes the same image. 9. A constant part calculating means for calculating a constant part which is used only for the reference image and which is determined only by the reference image when reconstructing the reference image, and stores the calculated constant part. Any one of 3,4,5,6,7,8
An image processing apparatus according to the item.