JP2001291105A - Method and device for recognizing pattern - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a new pattern recognizing method which can quickly calculate the position and attitude of a pattern and is hardly affected even if lighting and the reflection states of an object surface change. SOLUTION: In this method, a plurality of contour points of a template are detected to store a shift vector V in which the length (r) of a straight line connecting a contour point and a preliminarily defined reference point and the angle θ formed between the straight line and a normal vector at the contour point are defined as attributes (S1), contour points of an object pattern in an image are detected (S2 and S3) to calculate a pattern normal vector in a contour point of the object pattern (S4), shift vector conversion is performed for calculating the coordinates of a mapping point obtained by being shifted as much as the length (r) of the shift vector with respect to the pattern normal vector in the direction of the attribute angle θ (S5), the angle difference data between the pattern normal vector and a template normal vector are accumulated in a coordinate correspondence memory corresponding to the calculated mapping point coordinates (S6), an address where the number of pieces of data converges after the shift vector conversion is detected to obtain the object position on the basis of coordinates corresponding to the memory address (S8) and the object posture is obtained from the frequency distribution of angle difference data stored in the address (S9).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern recognition method and apparatus for identifying an object in an image and detecting its position and orientation.

[0002]

2. Description of the Related Art In an automatic operation, in particular, an automatic operation by a robot, it is an essential technique to recognize a position and a posture of an object by using an image of the object. Conventionally, as a method of recognizing a pattern in an image, a template matching method in which a template representing a pattern to be detected is applied to an image to calculate a cross-correlation value, and a position where this value is the largest is set as a position of an object is famous. is there. There is also a template matching method in which the sum of the absolute values of the differences is calculated and detected based on the position where the difference becomes the smallest.

However, in the template matching method, if there is a possibility that the orientation of the pattern is different from that of the target, the direction of the pattern is changed little by little until the target and the pattern match, and matching is performed from the end of the image. Because of the repetition, the amount of calculation becomes enormous and there is a disadvantage that it takes time. Further, when the measurement object is a metal product, the gray level of the object in the image greatly varies depending on the condition of the light beam, and therefore, when the difference from the outline of the template is large, the position of the object cannot be detected. Cases arise.

There is also a vector correlation method for detecting the position of an object from a result obtained by shifting and adding an image using a vector connecting the edge position of the template and a point to be detected, but when the posture of the object changes. Not applicable to

[0005]

An object of the present invention is to provide a new pattern recognition method capable of quickly calculating the position and orientation of a pattern. It is another object of the present invention to provide a pattern recognition method which is hardly affected by the above.

[0006]

In order to solve the above-mentioned problems, a pattern recognition method according to the present invention detects a plurality of contour points on a contour of a template image and a template normal vector at the contour point, and detects the contour. A shift vector V having attributes of a length r of a line segment connecting the point and a predetermined reference point and an angle θ between the line segment and a normal vector is created and stored in advance. Then, an image of the object is obtained, a contour point on the contour of the object pattern in the image is detected, a pattern normal vector at the contour point of the object pattern is calculated, and an attribute is assigned to the pattern normal vector. Performs a shift vector conversion for calculating the coordinates of a mapping point shifted by the length r of the shift vector in the direction of the angle θ, stores data in a coordinate corresponding memory corresponding to the calculated mapping point coordinates, and after the shift vector conversion. It is characterized in that an address where the number of data is concentrated is detected, and the position of the object is obtained based on coordinates corresponding to the memory address.

According to the pattern recognition method of the present invention, the comparison with the template is performed by a shift vector conversion mainly based on a simple repetitive operation, so that a high-speed operation is possible. Further, since the determination is performed by statistical means, the pattern can be accurately identified even if the quality of the acquired video is poor, and at the same time, the position of the object can be determined.

Further, by using the angle difference between the pattern normal vector and the template normal vector as data to be stored in the coordinate-corresponding memory, the frequency distribution of the angle difference values stored in the memory addresses corresponding to the coordinates of the object is obtained. , The posture of the target object can be obtained from, for example, a peak value or an average value. This is because the angle difference has substantially the same value for each mapping point for a pattern that matches the template, and means the angle between the directions of the orientation patterns of the template.

After the position of the object is obtained based on the memory address where the number of data is concentrated, the shift vector conversion is performed again, and the pattern method is used only when there is a mapping point corresponding to the vicinity of the obtained position. The angle difference between the line vector and the template normal vector may be stored and stored, and the orientation of the target object may be obtained based on the frequency distribution of the stored data values. In the method of obtaining the position and orientation at once, the angle difference is stored in the memory address corresponding to the coordinates of the mapping point, so that the memory structure needs to be three-dimensional and the memory capacity must be large. In the case of performing the shift vector conversion in two steps as described above, in order to record the mapping point position, it is sufficient to step forward the numerical value stored in the address corresponding to the mapping point coordinate, and accumulate the angle difference. Sometimes, it is sufficient to store only the angle difference data when the mapping point falls within a limited range, so that the memory capacity for storing the data may be small.

Further, after the position of the object is roughly obtained by the first shift vector conversion, further shift vector conversion is performed, and only when there is a mapping point corresponding to the vicinity of the obtained position, the coordinate correspondence is further finely divided. It is also possible to accumulate the appearance data of the mapping points in the memory and obtain a more precise position based on the concentration of the number of data. By combining the coarse measurement and the precise measurement in this way, it is possible to make the size of the grid corresponding to the coordinates large in the coarse measurement and to reduce the capacity of the coordinate-corresponding memory. Since only data corresponding to a narrow area near the data storage needs to be stored, measurement can be performed with sufficient accuracy even if the data storage memory is extremely small.

Instead of calculating the contour point from the contour of the gray image and calculating the normal vector at the contour point as described above, the distance to the object is recorded at the position corresponding to the pixel in the two-dimensional image memory. The position and orientation of the three-dimensional object can be obtained by performing a shift vector conversion by obtaining a normal vector of the object surface from the distance gradient at each point using the distance image and performing a shift vector conversion.

Further, in order to solve the above-mentioned problems, a pattern recognition device according to the present invention comprises an image processing device, an information processing device,
An image processing device that captures an image of a template, detects a contour point of the template and a normal vector at the contour point, and determines a contour point and a predetermined reference point. Shift vector V whose attributes are the length r of the connecting line segment and the angle θ between the line segment and the normal vector.
And the shift vector is stored in advance in the first storage device. At the time of measurement, the image processing apparatus inputs image information of the target object, and the target in the image is input in the same manner as in the template. An outline point is detected for an object pattern, a pattern normal vector at the outline point is calculated, and the information processing apparatus inputs the pattern normal vector and the shift vector to perform shift vector conversion.

The shift vector conversion is for calculating coordinates of a mapping point shifted by a line segment length r in a direction having the attribute angle θ of the shift vector with respect to the pattern normal vector. The second storage device includes a coordinate-corresponding memory having a memory cell corresponding to the coordinates, and counts each time the mapped point has the coordinates at a coordinate-corresponding memory address corresponding to the coordinates of the mapped point calculated by the information processing device. The frequency is stored by a method such as stepping, and the information processing apparatus reads the recording result after the shift vector conversion, examines the coordinates corresponding to the memory having the large frequency, and determines the position of the target based on the result. Features.

At first, the position of the object is roughly determined based on the memory in which the number of data is concentrated by performing the shift vector conversion using the coarsely-coordinated memory corresponding to the coordinates, and then further finely divided. A second coordinate-corresponding memory is prepared, the shift vector is converted again, and the appearance data of the mapping point is stored in the second coordinate-corresponding memory only when the mapping point comes near a roughly determined position. A more precise position may be obtained based on the degree of concentration of the stored data.

Instead of storing the frequency of appearance in the coordinate-corresponding memory, data of the angle difference between the pattern normal vector and the template normal vector is stored, and the coordinates of the memory in which the number of data is concentrated are stored in the memory. The position of the object may be determined, and the posture of the object may be determined based on the frequency distribution of the angle difference data values stored in the memory.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail based on embodiments with reference to the drawings.

[0017]

This embodiment shows a case where the position and orientation of a pattern in an image are detected by using the pattern recognition method of the present invention. FIG. 1 is a block diagram of the pattern recognition device of the present embodiment, FIG. 2 is a flowchart showing the pattern recognition method of the present embodiment, FIG. 3 is a flowchart of a process of creating a shift vector as preparation, and FIG. FIG. 5 is a flow chart of the shift vector conversion used in the present embodiment, FIGS. 6 to 9 are drawings for sequentially explaining the steps of the shift vector conversion, and FIGS.
FIG. 1 is a diagram for explaining conditions for determining the position of a pattern, FIG.
FIG. 1 is a conceptual diagram of a robot device to which the present embodiment is applied.

In this embodiment, a pattern recognition method which can judge a target product when the degree of approximation is high in comparison with the shape of the template and detect the position and orientation of the target product and which can be easily calculated by a computer. Then, as shown in FIG.
It is used in a pattern recognition device including an image processing device 1, an information processing device 2, a first storage device 3, and a second storage device 4. The pattern recognition procedure of the present embodiment is as shown in the flowchart of FIG. 2. First, an image pickup device (not shown) generates an image by taking an object to be sorted or a template of the shape of the object in a standard posture. The image processing apparatus 1 receives the image information, calculates a shift vector from the input image information, and stores it (S1).

FIG. 3 is a flowchart for explaining a method of creating a shift vector. The standard template TM is photographed by an imaging device having the same configuration as that used for measuring the target product, and image information as shown in FIG. 4 is obtained (S11). The standard template TM has the shape of the object to be identified,
It is placed in a standard posture. At this time, the information processing apparatus 2 cuts out the area where the template TM exists as a teaching area, and specifies a reference point O at an appropriate position (S12). The reference point O may be specified inside or outside the template TM.

The image processing apparatus 1 scans the teaching area of the grayscale image obtained by photographing the template TM with a contour extraction filter such as a Mexican hat filter, and extracts a contour part where a density gradient exists as a zero cross point Pi ( S13). At this time, the direction of the density gradient is also calculated and becomes the direction of the normal to the contour line (S1).
4). By scanning an appropriate filter in the image as described above, it is possible to mechanically calculate a point on the contour line and a normal vector set there.

The unit vector on the normal line set at the zero cross point Pi is called a teaching vector Ti (S15). The information processing device 2 receives information of the teaching vector Ti, generates a line segment from the zero-cross point Pi to the reference point O, and calculates the length thereof. A vector having the length ri of the line segment and the angle θi formed by the line segment with the teaching vector Ti as attributes (ri, θi) is called a shift vector Vi (S16). The obtained shift vector Vi (ri, θi) is stored in the first storage device 3.
(S17).

For the template TM in the image, the same procedure is repeated until the zero cross points Pi can be extracted at appropriate intervals over the entire circumference of the contour (S18). The number of contour points Pi employed on the outer periphery of the template TM may be determined according to the complexity of the shape of the template. However, since this method uses a statistical method for determination, so many points are used. No need. In this manner, the shift vectors V1 and V1 from the appropriate number of starting points Pi determined at appropriate intervals around the template TM to the reference point O are set.
V2,..., Vn are obtained.

After the shift vector Vi of the article to be identified is prepared in the first storage device 3, processing can be performed on the actual article to be identified. An image of a target article is photographed by an image pickup device (not shown), converted into a grayscale image, and taken into the image processing device 1 (S2). The target article is, for example, a product conveyed by a belt conveyor, or a semi-finished product on a carrier table and waiting for processing by a robot. The shooting of the target article is adjusted to the same scale as that of the template TM by adjusting the shooting conditions of the template TM.

The image processing apparatus 1 generates a gray image F of a target article.
, And a normal vector Nj at the detection point Qj is obtained (S3). The contour can be detected in the same manner as when the shift vector is generated, and at the same time, the direction of the density gradient at the detected contour point Qj, that is, the normal vector Nj can be calculated. Since this normal vector represents the inclination of the contour, it is also called a contour gradient vector.

By scanning the target area in the image with the contour extraction filter, the contour points Qj are detected at appropriate intervals over the entire circumference of the image F of the target article, and the normal vector Nj at that point is obtained. The information is stored in the second storage device 4 via the device 2. After finding the contour points Qj over the entire circumference of the target article pattern, the information processing device 2 performs shift vector conversion (S5).

The shift vector conversion is a procedure for obtaining a mapping point Mji obtained by applying a shift vector Vi to each point Qj as shown in the flowchart of FIG. FIG. 6 is an explanatory diagram showing a state in which an initial shift vector V1 is applied to an image including a first pattern F1 and a second pattern F2 to obtain a mapping point Mj1.

As shown in FIG. 6, each contour point Q1, Q2,
.., The shift vector V1 obtained around the template TM with Qm as a starting point is drawn (S21). At this time, the direction of the shift vector V1 is the normal vector N1,
N2,..., Nm form an attribute angle θ1, which is an attribute of each. Then, each point Q1, Q2,
.., A mapping point Mj1 moved by the shift vector V1 is obtained for each Qm. At this time, each point Q1, Q2,
.., N, normal vectors at Qm
An angle difference Φj1 between m and the teaching vector T1 of the shift vector V1 is calculated (S22).

The second storage device 4 is provided with an area having an address corresponding to a coordinate, that is, a coordinate-corresponding memory. The address has a three-dimensional structure including a plurality of memory elements capable of storing a plurality of data. ing. Mapping point Mj1
Is obtained, the angle difference Φj1 is recorded in the address corresponding to the coordinates of the mapping point Mj1 (S2).
3, S6).

As shown in FIG. 7, the second shift vector V
The same operation is performed for No. 2. FIG. 8 is a conceptual diagram in the case where processing is performed on the third shift vector V3.
FIG. 9 is a diagram showing the processing status of the last shift vector Vn. In this way, the above operation is performed for all the shift vectors V1, V2,..., Vn (S24, S7).

As described above, the shift vector conversion is completed for all the extracted contour points Qj by applying the entire shift vector Vi, and the mapping point Mji and the angle difference Φji
Is stored in the coordinate correspondence memory of the second storage device 4. FIG.
0 is a drawing in which the result of the shift vector conversion is overlaid on all the shift vectors. As shown in FIG. 10, the first pattern F among the patterns to be measured is
Since 1 has the same shape as the template TM, a mapping reference point Ot in which the mapping points Mji overlap by the number n of the shift vectors Vi is created. This mapping reference point Ot corresponds to the template T
A position corresponding to the reference point O of M, and the mapping reference point Ot
Can represent the position of the pattern F1.
Note that the mapping points related to the second pattern F2 different from the template TM are dispersed, and no concentrated mapping reference points appear.

As described above, when the same pattern as the template TM exists, an address in which the number of data is concentrated appears in the coordinate corresponding memory after the shift vector conversion is completed. The coordinates corresponding to this address become the position of the pattern (S8). However, the contour points Qj are extracted at appropriate intervals on the pattern contour, and are not necessarily located at positions completely corresponding to the contour points Pi on the template. Therefore, a statistical spread occurs at the coordinates where the mapping points are gathered. Therefore, it is preferable to set the cells of the coordinate correspondence memory to be slightly larger or to perform the position determination by gathering some adjacent addresses. The information processing device 2 reads out the relevant memory area and performs information processing to determine the position of the pattern.

The memory corresponding to the mapping reference point Ot also includes, as noise, angle difference Φji data calculated for a mapping point that does not match the template.
However, since the angle difference Φji has substantially the same value over the entire circumference of the contour for the mapping point Mji that matches the template, taking the frequency distribution of the stored angle difference Φji gives the angle difference Φji for the target pattern. Φji can be extracted. Pattern F1 targeted by this angle difference
(S9). In this way, the captured pattern can be compared with the template, the target pattern can be recognized, and its position and orientation can be obtained and output (S
10).

According to the pattern recognition method of this embodiment, the contour point and the inclination of the contour line at that point are obtained by simple image processing, and the shift vector conversion is mechanically performed by an extremely simple vector operation, and the Since the position information is obtained by majority decision of the number of data stored in the memory and the attitude is obtained from the frequency distribution of the data content, pattern recognition can be performed at a very high speed by a simple processing device. In addition, since the arithmetic processing is performed statistically, even if the illumination or the reflection state of the object surface slightly changes, the position and orientation can be reliably detected without being affected by the change.

The number or interval of contour points to be used on the outer periphery of the template and the pattern may be determined according to the complexity of the shape of the target article or pattern. However, since a statistical method is used, There is no need to use as many points. Further, it is convenient in arithmetic processing to use a number such as 16 or 32 according to the number of arithmetic bits of the computer central processing unit.

Further, in the above embodiment, the accumulation of the number of mapping points and the accumulation of the angle difference data are performed simultaneously. However, only the mapping points are calculated by the first shift vector conversion and stored in the coordinate correspondence memory to store the mapping points. Pattern matching is performed based on the degree of concentration, and then shift vector conversion is performed again.This time, an appropriate area surrounding the mapping reference point where the mapping points are concentrated is designated, and the angle difference data is stored only in this area. The posture may be detected from the frequency distribution. According to such a two-pass method, there is no need to prepare a memory having a three-dimensional structure for storing angle difference data at all the coordinate corresponding addresses, and the first mapping point data is stored as the number of data in the coordinate corresponding memory. Since the later angle difference data only needs to be selected and stored for a limited address, the memory capacity can be saved.

In the above-described embodiment, the shift vector is applied to the contour points all around the pattern at the time of the shift vector conversion. However, the order is reversed so that the contour points are shifted one by one. All the vectors may be operated. Obviously, the results obtained are the same, and when monitoring the degree of data integration, a sufficiently large amount of data can be stored in the same memory corresponding to coordinates during shift vector conversion, and the coordinates of the object can be known. Sometimes you can. In such a case, when it is found that the accuracy is sufficiently high, it is possible to switch to the precision measurement in which the address is limited to obtain precise position information.

As another aspect of the present embodiment, there is a method of detecting a three-dimensional position and orientation using a distance image. The distance image is a two-dimensional image in which distance information to the object surface is stored at each point corresponding to a pixel in the two-dimensional memory space. The distance image is obtained by, for example, scanning a target with a laser beam and forming an image of the reflected light on a CCD by a lens.
It can be generated by a method of using the light spot position information of a CD, converting the distance into a distance based on the principle of triangulation, and storing the distance in an image memory. In such a distance image, when a distance gradient near the target point is obtained, a normal vector of the object surface can be obtained.

With respect to the reference solid corresponding to the template,
Generate normal vectors on the surface at appropriate intervals. This normal vector is a three-dimensional vector called a teaching vector. Next, a shift vector from the foot of the teaching vector toward the reference point is generated, and the length between the shift vector and the angle between the teaching vector and the direction of the teaching vector are recorded as attributes.

After the preparation for measurement is completed in this way,
A distance image is acquired for a target object. The distance image information is taken into the image processing apparatus, and the normal vector of the object surface is calculated at an appropriate interval from the distance gradient. A shift vector conversion is performed for each normal vector. In this embodiment, since the shift vector is a three-dimensional vector, the mapping points are distributed in a three-dimensional space. Therefore, a memory having a three-dimensional structure is selected as the coordinate correspondence memory for storing the coordinates of the mapping point. The coordinate correspondence memory stores a solid angle between the teaching vector and the normal vector of the target object.

If there are coordinates where the mapping points are concentrated, the figure matches the template. When the distribution of the solid angles stored in the memory corresponding to the coordinates where the mapping points are concentrated is examined to find the angle having the highest frequency, this represents the posture of the target object. When a part different from the template is photographed depending on the attitude of the three-dimensional figure, the concentration of the mapping points is smaller than the number of the shift vectors, and thus the identity may be determined based on an appropriate number of concentrations.

The pattern recognition method of the present invention can be applied to a robot device as shown in FIG. The semi-finished product 12 conveyed by the conveyor 11 is photographed by the imaging device 13 and image information is sent to the image processing device 14. The image processing device 14 detects the contour of the captured pattern by filtering the grayscale image and calculates a normal vector on the contour. The calculation result is stored in the storage device 16 via the central processing unit 15.

The central processing unit 15 executes shift vector conversion using the teaching vector and the shift vector relating to the template stored in the storage device 16 in advance.
The obtained angle difference data is sequentially stored in a coordinate correspondence memory in the storage device 16. When the shift vector conversion is completed, an address where the data in the coordinate correspondence memory is concentrated is found, and the frequency distribution of the angle difference stored at the address is examined. The detected address is used as the position of the pattern,
When the obtained information is sent to the robot controller 17 with the most frequent angle difference value as the pattern posture, the robot controller 17 operates the robot hand 18 to operate the semi-finished product 12.
And accurately adjust the posture relationship and transport it to the target position.

[0043]

As described above, according to the pattern recognition method of the present invention, a required object is accurately and quickly identified from an image of a product or the like, and its position and orientation are calculated. Information can be transmitted to a robot controller or the like, and a robot hand or the like can be automatically operated to grasp the item.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment of a pattern recognition device according to the present invention.

FIG. 2 is a flowchart illustrating an embodiment of a pattern recognition method according to the present invention.

FIG. 3 is a flowchart of a process of creating a shift vector in the embodiment.

FIG. 4 is a diagram illustrating a method for generating a shift vector in the present embodiment.

FIG. 5 is a flowchart of shift vector conversion used in the present embodiment.

FIG. 6 is a diagram illustrating a first step of the shift vector conversion according to the present embodiment.

FIG. 7 is a diagram illustrating a second step of the shift vector conversion according to the present embodiment.

FIG. 8 is a diagram illustrating a third step of the shift vector conversion according to the present embodiment.

FIG. 9 is a diagram illustrating a fourth step of the shift vector conversion of the present embodiment.

FIG. 10 is a diagram illustrating conditions for determining a position of a pattern in the present embodiment.

FIG. 11 is a conceptual diagram of a robot device to which the present embodiment is applied.

[Explanation of symbols]

 14 Image processing device 15 Central processing device 16 Storage device 17 Robot control device

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B057 DA07 DC08 DC16 DC30 DC33 DC36 5L096 FA06 FA14 FA67 FA69 JA09

Claims (11)

[Claims] 1. A method for detecting a normal vector at a plurality of points on an object surface from an image of a template, the length r of a line segment connecting the surface point and a predetermined reference point, and the angle between the line segment and the normal vector. A shift vector V having θ as an attribute is created and stored in advance, an image of the object is obtained, a pattern normal vector at a surface point of the object pattern in the image is calculated, and the pattern normal vector The attribute angle θ
The shift vector conversion for calculating the coordinates of the mapping point shifted by the line segment length r in the direction having the following is performed, the data is stored in the coordinate correspondence memory corresponding to the mapping point coordinates, and the number of data after the shift vector conversion is calculated. A pattern recognition method characterized in that a position of an object is obtained based on coordinates corresponding to a memory in which is concentrated. 2. The pattern recognition according to claim 1, wherein the image of the template is a grayscale image, and the normal vector of the surface of the object is a normal vector at a contour point detected from the grayscale image. Method. 3. The image according to claim 1, wherein the image of the template is a distance image, and the normal vector of the object surface is a normal vector calculated for a point on the object surface in the distance image. Pattern recognition method. 4. A plurality of contour points on the contour and a normal vector at the contour point are detected from an image of the template, and a length r of a line segment connecting the contour point and a predetermined reference point, and the line segment are determined. A shift vector V having the angle θ of the normal vector as an attribute is created and stored in advance, an image of the object is acquired, and a contour point on the contour of the object pattern in the image is detected. Calculate the pattern normal vector at the contour point of the object pattern, perform shift vector conversion to calculate the coordinates of the mapping point shifted by the line segment length r in the direction having the attribute angle θ from the pattern normal vector, A pattern storing data in a coordinate-corresponding memory corresponding to the mapping point coordinates, and obtaining a position of an object based on coordinates corresponding to a memory in which the number of data is concentrated after shift vector conversion.識方 method. 5. The method according to claim 1, wherein after determining the position of the object based on the memory in which the number of data is concentrated, further performing a shift vector conversion and having a mapping point corresponding to the vicinity of the determined position. 5. The method according to claim 1, wherein an angle difference between a normal vector and a normal vector of the template is stored and stored, and a posture of the object is obtained based on a frequency distribution of the stored data values. The pattern recognition method according to any one of the above. 6. After the position of an object is determined based on the memory in which the number of data is concentrated, shift vector conversion is further performed to further fine-tune the case where there is a mapping point corresponding to the vicinity of the determined position. 6. The data according to claim 1, wherein the appearance data of the mapping point is stored in the divided coordinate correspondence memory, and a more precise position is obtained based on the degree of concentration of the stored data. Pattern recognition method. 7. The method according to claim 7, wherein the data stored in the memory is an angle difference between the pattern normal vector and the template normal vector, and the target is based on coordinates corresponding to the memory in which the number of data is concentrated. 5. The position according to claim 1, wherein the position of the object is obtained, and the attitude of the object is obtained based on a frequency distribution of data values stored in a memory in which the number of data is concentrated. Pattern recognition method. 8. An image processing device, an information processing device, a first storage device, and a second storage device, wherein the image processing device extracts a plurality of contour points on a contour from a template image and a normal vector at the contour point. And a shift vector V whose attributes are the length r of the line segment connecting the contour point and the predetermined reference point and the angle θ between the line segment and the normal vector are created. The shift vector is stored in advance, and the image processing apparatus inputs image information of an object, detects an outline point on an outline of the object pattern in the image, and detects an outline point on the outline of the object pattern. A pattern normal vector is calculated, and the information processing apparatus inputs the pattern normal vector from the image processing apparatus, inputs the shift vector from the first storage device, and calculates the shift vector of the shift vector with respect to the normal vector. A shift vector conversion for calculating coordinates of a mapping point shifted by the line segment length r in a direction having the sex angle θ, and the second storage device includes a coordinate correspondence memory having a storage cell corresponding to the coordinates. The frequency at which the mapping point has the coordinates is stored in a coordinate correspondence memory corresponding to the calculated mapping point coordinates, and the information processing apparatus converts the shift vector after the shift vector conversion and stores the frequency based on the coordinates corresponding to the memory having the higher frequency. A pattern recognition device for determining the position of a pattern. 9. An image processed by the image processing apparatus is a distance image, wherein the contour point is a point on a surface of an object, and a normal vector is obtained by a distance gradient derived from the distance image. The pattern recognition device according to claim 8, wherein 10. After the position of the object is determined based on the memory in which the number of data is concentrated, shift vector conversion is further performed to further finely map the data when there is a mapping point corresponding to the vicinity of the determined position. 10. The pattern according to claim 8, wherein appearance data of the mapping point is stored in the divided second coordinate correspondence memory, and a more precise position is obtained based on the degree of concentration of the stored data. Recognition device. 11. A memory in which, instead of storing the frequency in the coordinate-corresponding memory, data of an angle difference between the pattern normal vector and the template normal vector is accumulated, and the data number is concentrated. The pattern recognition apparatus according to claim 8, wherein an orientation of the object is obtained based on a frequency distribution of the angle difference data values stored in the object.