JP2001175865A - Image processing method and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing method and its device which can perform a difference process corresponding to the importance of inspection and make an accurate nondefective/defective decision. SOLUTION: When a program stored in a program storage memory 4 is executed, a CPU 6 performs a difference process for calculating a luminance difference by putting a previously registered instruction image over a picked-up image of an object of inspection obtained by a TV camera 1, and performs multiplication by previously registered values of weight (deviation data) by pixels of the instruction image of the object to recalculate a difference value. The recalculated value is compared with a threshold to judge whether the object of inspection is nondefective or defective.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method and apparatus for performing inspection processing on an image obtained by imaging an object to be inspected and determining whether the object is good or defective. .

[0002]

2. Description of the Related Art When inspecting the quality of an inspection object (object work) by using an image processing apparatus, an image pattern (teaching image) of the object (work) that has been climbed in advance is inspected. Is superimposed on the inspection image obtained by imaging
The absolute value of the luminance difference between the two is calculated for each corresponding pixel, the total area of the region exceeding the luminance difference threshold separately set by the operator is calculated, and the area value is compared with the difference area threshold to determine whether the area is good or bad. Conventionally, a method of performing the defect has been adopted.

[0003]

In the case of the above-mentioned conventional method, the above-mentioned luminance difference threshold value is set to a constant value for all pixels in the teaching image, regardless of the importance for good / bad judgment. , The total area of all pixels exceeding the difference threshold has been calculated. Further, the brightness of an image near the contour of the object slightly changes due to the instability of the imaging system, and there is a problem that a difference in this portion is likely to occur even in a processing result for a good product.

[0004] In order to solve this problem, there is a method of performing a masking process on the above-mentioned portion to completely remove the portion from the inspection target. However, this area is an important portion as an inspection portion.
It was necessary to output a difference for a large difference in the outline shape, and it could not be completely excluded from the processing target.

[0005] In addition, depending on the object, the importance of the judgment is different depending on the part, and it is necessary to recognize a part having a high importance as a difference even if a slight difference is obtained. In some cases, it is desirable not to recognize the difference as a difference.

However, in the conventional method, since a fixed threshold value is set for all pixels, it is difficult or impossible for an operator to determine a difference threshold value and a difference area threshold value that satisfy the above-described conditions.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and has as its object the case where the contours are subtly changed due to the difference in importance between portions in a teaching image or the influence of an imaging system. Even in such a case, it is an object of the present invention to provide an image processing method and an image processing method capable of removing the influence thereof, executing a difference process according to the degree of importance in inspection, and accurately determining good / bad.

[0008]

According to the first aspect of the present invention, a difference in luminance between each pixel is calculated between a teaching image of an object which has been climbed in advance and an inspection image obtained by imaging an inspection object. In an image processing method of performing a difference process and outputting a total area of pixels whose difference result obtained by calculation exceeds a preset difference threshold, a weight is set for each pixel of the teaching image, For a low pixel, the difference result is made equal to the difference threshold, and the area of the pixel is not counted as the area of the pixel exceeding the difference threshold, so that the difference is not recognized as a difference.

According to a second aspect of the present invention, in the first aspect of the present invention, the setting of the weight for each pixel of the teaching image includes the teaching image and a plurality of picked-up images of the inspection object known to be good. , And for pixels exceeding the difference threshold, the weight is automatically calculated in a range of 0 to 1.0 so that the difference result is equal to or smaller than the threshold. .

According to a third aspect of the present invention, in the first aspect of the present invention, the setting of the weight for each pixel of the teaching image is manually performed.

According to a fourth aspect of the present invention, in any one of the first to third aspects of the present invention, before the execution of the difference processing, a pattern matching process of the captured image is performed by using the teaching image in advance. Detecting a plurality of candidates for a position and a rotation angle at which the degree of coincidence obtained by the processing exceeds a predetermined threshold, performing the above-described difference processing on each of the plurality of candidates as a target of the difference processing; I do.

According to a fifth aspect of the present invention, in any one of the first to third aspects of the present invention, before the execution of the difference processing, a pattern matching process of the captured image is performed by using the teaching image in advance. A plurality of positions and rotation angles at which the degree of coincidence obtained by the processing exceeds a predetermined threshold are detected, and the above-described difference processing is performed while changing the positions and rotation angles in the vicinity of the positions of the plurality of candidates, and each candidate is detected. It is characterized in that a result when the difference area value becomes minimum for each position is output as a final result.

According to the sixth aspect of the present invention, a difference process for calculating a difference in luminance between each pixel between a captured image of the inspection object by the TV camera and a teaching image of the object registered in the memory in advance is performed. Processing means, the plurality of detected candidates are set as target portions of the difference processing, and the difference processing is performed on each of the target portions, and a difference result obtained by calculation is set to a predetermined difference threshold value. The image processing apparatus for outputting the total area of the exceeded pixels includes a means for setting a weight for each pixel of the teaching image, and performing a process of making a difference result equal to a difference threshold value for a pixel of low importance. Preparing the detected plurality of candidates as target positions for the difference processing and executing the above-described difference processing on each target portion, and recognizes the area of the pixel as a difference without counting the area of the pixel exceeding the difference threshold. do not do And wherein the Unisuru.

According to a seventh aspect of the present invention, in the sixth aspect of the present invention, the setting of the weight for each pixel of the teaching image includes the teaching image and a plurality of picked-up images of the inspection object known to be good. The difference processing is performed between pixels, and for a pixel exceeding the difference threshold, the weight is automatically calculated in a range of 0 to 1.0 so that the difference result is equal to or less than the threshold, and the calculation is performed by calculation means. I do.

According to an eighth aspect of the present invention, in the sixth aspect of the present invention, the setting of the weight for each pixel of the teaching image is an input setting by manual operation means.

According to a ninth aspect of the present invention, in any one of the sixth to eighth aspects, a pattern matching process of the captured image is performed in advance using the teaching image before the execution of the difference process. A means for automatically detecting a plurality of candidates for a position and a rotation angle at which the degree of coincidence obtained by the processing exceeds a predetermined threshold value, and using these detected plurality of candidates as target places for difference processing, for each target place The difference processing is performed.

According to a tenth aspect of the present invention, in any one of the sixth to eighth aspects of the present invention, before the execution of the difference processing, a pattern matching process of the captured image is performed in advance by using the teaching image to perform pattern matching. Means for automatically detecting a plurality of candidates for the position and the rotation angle at which the degree of coincidence obtained by the processing exceeds a predetermined threshold value, and performing the difference processing while changing the position and the rotation angle in a region near the position of the detected plurality of candidates. And outputting a result when the difference area value becomes minimum for each position of each candidate as a final result.

[0018]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to embodiments.

(Embodiment 1) FIG. 1 shows an overall configuration of an image processing apparatus according to the present invention. The apparatus comprises a TV camera 1 for imaging an inspection object, and analog image data obtained by imaging. A / D converter 2 for converting image data into digital image data, image memory 3 for storing image data digitized by A / D converter 2, and a program storing a program for an image processing method and data for image processing Memory 4, a main memory 5 for executing processing, a CPU 6 for executing processing such as image processing, a monitor 7 for displaying a captured image of an inspection object, a processed image during image processing, and the like, and a monitor 7 D / A conversion for obtaining analog image data to be projected on
/ A converter 8, TV camera 1 to image memory 3
And a controller PLD for controlling connection from the image memory 3 to the monitor 8 via the D / A converter 7 and from the CPU 6 to the image memory 3. The connection from the TV camera 1 to the image memory 3, from the image memory 3 to the monitor 8 via the D / A converter 7, and from the CPU 6 to the image memory 3 is controlled so that two or more connections are not made simultaneously.

When the apparatus is started, the CPU 6
Reads the program and the data for processing of the image processing method of the present embodiment from the program storage memory 4 and executes the processing of the image processing method of the present embodiment on the main memory 5.

Next, an image processing method according to the present embodiment will be described.

First, in the present embodiment, a weight (deviation data) is set for each pixel of the teaching image of the object, and the luminance difference between the teaching image and the captured image of the inspection object is multiplied by the value of the weight. The difference value is recalculated, and a process of comparing the calculation result with the threshold value is performed. Here, the weight is set to a small value for a pixel that can accept a luminance difference with a low importance or a large luminance, thereby reducing the sensitivity to the luminance difference and making it difficult to recognize the difference.

For example, as shown in FIG. 2A, an image A 'of the inspection object W' is taken as an image of the inspection object W 'with respect to the teaching image A of the object W as a reference. (B)
As shown in FIG. 2, if the teaching image A is thinner than the image of the target object W, the user does not want to judge that the object is defective.
As shown in (d), when it is desired to recognize only the foreign matter X outside the inspection object W as a difference, a process is performed in which the luminance difference near the contour is not regarded as a difference.

That is, a weight value (deviation data) is calculated for each pixel with respect to the result obtained by superimposing the captured image A 'of the inspection object W' on the teaching image A and calculating the luminance difference of each pixel. The part that you do not want to recognize as the result of multiplication and difference (Fig. 2
The CPU 6 performs a correction process on the difference result so that the contour portion of the inspection object W ′ as in (b) becomes equal to (or smaller than) the threshold value.

The calculation formula at this time is as the following formula (1).

Diff (x, y) = | f (x, y) −g (x, y) | × h (x, y) (1) diff (x, y): coordinates (x, y) F (x, y): Inspection image luminance value at coordinates (x, y) g (x, y): Teach image luminance value at coordinates (x, y) h (x, y): Coordinates (x, y) y) Deviation data where coordinates are in pixel units. FIG. 3A shows an area where the luminance value fluctuates in the outline portion (black portion), and FIG. 3B shows a setting state of the deviation data. In the illustrated example, 0.75 is set for the contour portion of the image of the target object, and 1.0 is set for other portions.

Now, the processing procedure for the picked-up image A 'of the inspection object W' in FIG. 2B will be described with reference to FIG.

First, a captured image A '(corresponding to the image A' in FIG. 2B) of the inspection object W 'obtained as shown in FIG. 4A and a teaching image A (FIG. 2A) 4 (b), a large luminance difference appears in the outline portion (the portion painted black), and the value of this luminance difference is calculated by the difference processing. If the difference result is 128, the difference result exceeds the threshold if the difference threshold value for good / bad judgment is, for example, 96. However, in the processing of the present embodiment, the above-described deviation data is calculated for each pixel. , The deviation data value of 0.75
Is multiplied, and the result of the multiplication is compared with a threshold value (FIG. 4C). Here, since 128 × 0.75≤threshold, an image obtained by binarization based on the threshold is shown in FIG.
As shown in (d), the inspection is determined to be (non-defective).

Next, the processing procedure for the picked-up image A 'of the inspection object W' in FIG. 2C will be described with reference to FIG.

First, the captured image A ′ of the inspection object W ′ (corresponding to the image A ′ of the object in FIG. 2C) obtained as shown in FIG. 5A and the teaching image A are superimposed. 5 (b), a large luminance difference appears between the contour portion and the portion of the foreign matter X (the portion painted black), and the value of the luminance difference is calculated by the difference processing. If each of the difference results is 128, it exceeds the difference threshold value of 96. However, in the processing of this embodiment, since the above-described deviation data is set for each pixel, the difference result (128) Is multiplied by the deviation data value 0.75, the difference result (128) of the portion of the foreign matter X is multiplied by the deviation data value 1, and the result of the multiplication is compared with a threshold value (FIG. 5C )). Here, 128 × 0.75≤threshold in the contour portion, but 128 × 1 > threshold in the portion of the foreign matter X. Therefore, an image obtained by binarization based on the threshold value is shown in FIG. As a result, the foreign matter X portion is extracted, the area of the extracted pixel is counted, and the pass / fail is determined by comparing the count value with the difference area value. In this case, since it is larger than the difference area value set in advance, it is determined to be (defective).

Further, the processing procedure for the picked-up image A 'of the inspection object W' in FIG. 2D will be described with reference to FIG.

First, a captured image A ′ of the inspection object W ′ (corresponding to the image A ′ of the object in FIG. 2C) obtained as shown in FIG. And the position of the contour of the inspection object W 'in the captured image A' matches the position of the contour of the object W in the teaching image A, so that the luminance difference between the two is 0, but the foreign matter X is captured in the captured image A ' Therefore, as shown in FIG. 6B, a large luminance difference appears in the portion of the foreign matter X (a portion painted black), and the value of the luminance difference is calculated by the difference processing.
If the difference result is 128, even if multiplied by the value 1 of the deviation data, it exceeds the difference threshold value of 96 (FIG. 6).
(C)). Therefore, even if multiplied by the value of the deviation data, the image obtained by binarization based on the difference threshold value results in the extraction of the foreign matter X portion as shown in FIG. 5D. Good).

In the present embodiment, the above-described deviation data for each pixel is automatically determined by calculation processing of the CPU 6. In other words, the inspection object W 'that is known to be good
Is performed between the plurality of captured images A ′ and the teaching image A, and weighting is performed on the pixels exceeding the difference threshold so that the luminance difference value becomes equal to (or smaller than) the threshold value. Automatic adjustment within the range of 0 to 1.0.

That is, first, all the deviation data (h_o1d) are set to 1.0, and the difference processing is executed. Based on this execution result, the deviation data (h_new) is calculated by the following equation (2) for the pixel that has become “1” in the binarization processing using the difference threshold.
FIG. 7 shows the result. FIG. 7A shows a portion which has become “1” in the binarization process using a difference threshold by blacking it out, and FIG. 7B shows a calculation result of the deviation data. The value is 1 and the value in the shaded area is 0
75 is set.

Diff (x, y) = | f (x, y) −g (x, y) | × h_o1d (x, y) h_new (x, y) = threshold / diff (x, y) if diff (x, y)> threshold (2) h_new (x, y) = h_old (x, y) othewrwise h_new (x, y): Deviation data of newly calculated coordinates (x, y) h_old (x, y): Deviation data of previous coordinates (x, y) diff (x, y): Difference result of coordinates (x, y) threshold: Difference threshold f (x, y): Inspection image of coordinates (x, y) Luminance value g (x, y): luminance value of the teaching image at coordinates (X, y) where the coordinates are pixel units. From the second time on, the difference value is calculated based on the previous deviation data.
The processing of equation (2) is performed. This process is repeated for the prepared image A '. When the inspection object W ′ that cannot be sufficiently dealt with by the set weight appears while the inspection is continued, the same processing is performed to further correct the deviation data.

As described above, in this embodiment, the value of the deviation data is automatically set for each pixel.

(Embodiment 2) In the case of Embodiment 1, in automatically setting the value of the deviation data, a plurality of inspection objects must be prepared. In addition, when the part where the value of the deviation data is changed is known in advance, extra time is required.

Therefore, in this embodiment, the operator handling the image processing apparatus directly sets the weight of each pixel. Accordingly, when there is no need to prepare a plurality of captured images before setting and the difference in importance between pixels is clear, or when a skilled operator operates, the work at the time of setting can be reduced. . In the case where an inspection object that cannot be sufficiently dealt with with the set weight appears during the continuation of the inspection, the processing of the first embodiment can be performed to further correct the deviation data.

In this embodiment, it is needless to say that a keyboard, a pointing device and the like are provided in the image processing device as means for the operator to perform the weighting operation, and the data of the input operation is processed by the CPU 6.

(Embodiment 3) In the difference processing of Embodiments 1 and 2, the position and the rotation angle at which the teaching image and the inspection image are superimposed must be determined by some method. Therefore, in the present embodiment, the position and angle are detected by the pattern matching processing using the teaching image A used for the difference processing in the first and second embodiments. When a program is set to provide the CPU 6 with a function of automatically detecting a plurality of locations that exceed a (threshold for pattern matching) in the CPU 6 and the apparatus is started, the CPU 6 sets a plurality of candidates for locations where the degree of coincidence exceeds the threshold by the function. Is automatically detected, and after the automatic detection, the processing of the first embodiment (or the second embodiment) is performed on the plurality of locations.

That is, when there are a plurality of inspection objects W 'to be inspected in the picked-up image A', a plurality of positions and angles whose coincidence is equal to or larger than the threshold are detected by the pattern matching processing using the same teaching image A as the difference processing. By superimposing the teaching image A on each position and angle, the difference processing described in the first embodiment is executed.

By performing the difference processing on a plurality of position / rotation angle candidates in this manner, the preprocessing such as reducing the number of inspection objects W ′ in the captured image A ′ to one becomes unnecessary.
When there are a plurality of objects to be inspected, such as a quality inspection of a print character string, processing can be performed at once, and the processing time of the entire image processing apparatus can be reduced.

Further, since the rotation angle of the inspection object W 'is also detected by pattern matching, the inspection object W' is allowed to rotate arbitrarily. Need not be added, and the apparatus and method can be simplified.

Further, by detecting the position and the rotation angle by the pattern matching processing, the inspection object largely separated from the object of the teaching image A can be determined to be defective before executing the difference processing. You can save time. In the pattern matching processing, first, the teaching image A is raster-scanned from the upper left end to the lower right end of the captured image A ', the degree of coincidence is calculated at each position, and a plurality of images having a threshold or more are selected as candidates. The teaching image A is rotated at each selected position, and the degree of coincidence is calculated for each angle. The angle at which the degree of coincidence is maximum is defined as the rotation angle at that position. At this point, those whose degree of coincidence is less than the threshold are excluded from the candidates. The difference processing described in the first embodiment is performed on the position / rotation angle.

(Embodiment 4) In this embodiment, the pattern matching processing using the teaching image A used for the difference processing in Embodiment 1 (or Embodiment 2) as a template makes the matching degree a threshold (pattern matching threshold). , A plurality of candidates are detected for the position and the rotation angle exceeding the range, and the difference processing of the first embodiment (or the second embodiment) is executed while changing the position and the rotation angle in the vicinity of the detected position.
The result when the difference area value becomes minimum for each candidate position is output as the final result.

The inspection object W 'of the picked-up image A' may have some difference from the teaching image A, so that its position and angle may not be accurately detected only by the pattern matching processing. . If the difference is large, an unnecessary difference is output when the difference processing is actually performed, and there is a risk that a non-defective product is erroneously determined to be defective.

Accordingly, the position / rotation angle detected by the pattern matching is not used as it is, but the operator designates a position / angle range in the vicinity thereof, executes the difference processing within the designated range, and obtains the difference area value. The processing result for the position and rotation angle at which is minimized is regarded as the processing result for that position and rotation angle.

Thus, the position of pattern matching
The erroneous determination of good / bad due to the rotation angle detection error can be reduced.

[0049]

According to the first aspect of the present invention, a difference process is performed for calculating a difference in luminance between each pixel of a teaching image of an object which has been climbed in advance and an image of an inspection object. In an image processing method for outputting the total area of pixels in which a difference result obtained by calculation exceeds a preset difference threshold, a weight is set for each pixel of the teaching image, and a pixel having a low importance is set. In other words, the difference result is made equal to the difference threshold, and the area of the pixel is not counted as the area of the pixel exceeding the difference threshold and is not recognized as a difference. In addition, even when the contour slightly changes due to the influence of the imaging system, the influence is removed, and the difference processing according to the degree of importance in the inspection can be executed.

According to a second aspect of the present invention, in the first aspect of the present invention, the setting of the weight for each pixel of the teaching image includes the teaching image and a plurality of picked-up images of the inspection object known to be good. 2. A difference process is performed between pixels and a pixel having a difference threshold value is automatically calculated in a range of 0 to 1.0 so that a difference result is equal to or smaller than the threshold value. The weight can be automatically set in addition to the effect of the present invention.

According to a third aspect of the present invention, in the first aspect of the present invention, the weight is set manually for each pixel of the teaching image, so that it is not necessary to prepare a plurality of captured images before setting. When the difference in importance is clear or when a skilled operator operates, there is an effect that the work at the time of setting can be reduced.

According to a fourth aspect of the present invention, in any one of the first and second aspects of the present invention, a pattern matching process of the picked-up image is performed by using the teaching image in advance before the execution of the difference processing. The position and the rotation angle at which the degree of coincidence obtained by the processing exceeds a predetermined threshold are detected as a plurality of candidates, and the plurality of candidates are subjected to difference processing as target portions.
Since the above-described difference processing is executed for each target location, there is an effect that restrictions on pre-processing and devices can be reduced, and the entire processing can be speeded up.

According to a fifth aspect of the present invention, in any one of the first to third aspects of the present invention, before the execution of the difference processing, a pattern matching process of the captured image is performed in advance by using the teaching image to perform pattern matching. A plurality of positions and rotation angles at which the degree of coincidence obtained by the processing exceeds a predetermined threshold are detected, and the above-described difference processing is performed while changing the positions and rotation angles in the vicinity of the positions of the plurality of candidates, and each candidate is detected. Since the result when the difference area value is minimized for each position is output as the final result, the influence of the position / rotation angle detection error of pattern matching due to the difference in the work between the teaching image and the captured image is eliminated, and the pass / fail is determined. There is an effect that erroneous determination of can be avoided.

According to a sixth aspect of the present invention, a difference process for calculating a difference in luminance between each pixel between a captured image of an inspection object by a TV camera and a teaching image of the object registered in a memory in advance is performed. Processing means, the plurality of detected candidates are set as target portions of the difference processing, and the difference processing is performed on each of the target portions, and a difference result obtained by calculation is set to a predetermined difference threshold value. The image processing apparatus for outputting the total area of the exceeded pixels includes a means for setting a weight for each pixel of the teaching image, and performing a process of making a difference result equal to a difference threshold value for a pixel of low importance. Preparing the detected plurality of candidates as target positions for the difference processing and executing the above-described difference processing on each target portion, and recognizes the area of the pixel as a difference without counting the area of the pixel exceeding the difference threshold. I will not do it Because it is possible to realize a device which effects of the invention of claim 1 is obtained.

According to a seventh aspect of the present invention, in the sixth aspect of the present invention, the setting of the weight for each pixel of the teaching image includes the teaching image and a plurality of picked-up images of the inspection object known to be good. Difference processing is performed between pixels, and for a pixel exceeding a difference threshold, calculation is performed by a calculation unit that automatically calculates a weight in a range of 0 to 1.0 so that the difference result is equal to or smaller than the threshold. A device that can achieve the effects of the second invention can be realized.

According to an eighth aspect of the present invention, in the sixth aspect of the invention, since the setting of the weight for each pixel of the teaching image is an input setting by manual operation means, the effect of the third aspect of the invention is obtained. The device can be realized.

According to a ninth aspect of the present invention, in the invention according to any one of the sixth to eighth aspects, before executing the difference processing, a pattern matching process of the captured image is performed by using the teaching image in advance. A means for automatically detecting a plurality of candidates for a position and a rotation angle at which the degree of coincidence obtained by the matching process exceeds a predetermined threshold is provided. Since the above-described difference processing is executed in this way, it is possible to realize an apparatus that can achieve the effects of the invention of claim 4.

According to a tenth aspect of the present invention, in accordance with any one of the sixth to eighth aspects, the pattern matching of the captured image is performed in advance using the teaching image before the execution of the difference processing. Means for automatically detecting a plurality of candidates for the position and the rotation angle at which the degree of coincidence obtained by the processing exceeds a predetermined threshold value, and performing the difference processing while changing the position and the rotation angle in a region near the position of the detected plurality of candidates. And a means for outputting, as a final result, a result when the difference area value is minimized for each position of each candidate, so that an apparatus that can achieve the effects of the invention of claim 5 can be realized.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an image processing apparatus used in the present invention.

FIG. 2A is a diagram illustrating an example of a teaching image used in the first embodiment of the present invention. FIG. 3B is a diagram illustrating an example of a captured image of the inspection target object determined to be non-defective in the above. FIG. 3C is a diagram illustrating an example of a captured image of an inspection target object that is determined to be defective in the above.
(D) is another example of a picked-up image of the inspection object to be determined to be defective in the above.

FIG. 3A is an explanatory diagram of the above. (B)
FIG. 3 is a setting diagram of deviation data used in the embodiment.

FIG. 4 is an explanatory diagram of an inspection process according to the first embodiment.

FIG. 5 is an explanatory diagram of an inspection process according to the embodiment.

FIG. 6 is an explanatory diagram of an inspection process according to the embodiment.

FIG. 7 is an explanatory diagram of automatic setting of weights according to the embodiment.

[Explanation of symbols]

 Reference Signs List 1 TV camera 2 A / D converter 3 Image memory 4 Program storage memory 5 Main memory 6 CPU 7 Monitor 8 D / A converter PLD controller

Claims (10)

[Claims] 1. A difference process for calculating a difference in luminance between pixels between a pre-registered teaching image of a target object and a picked-up image obtained by picking up an image of an inspection target object, and calculating the difference. In an image processing method for outputting a total area of pixels whose difference results exceed a preset difference threshold, a weight is set for each pixel of the teaching image, and the difference result is subtracted for pixels of low importance. An image processing method characterized in that an area of the pixel is not counted as an area of a pixel exceeding a difference threshold and is not recognized as a difference by making the area of the pixel equal to a threshold. 2. A method for setting a weight for each pixel of the teaching image, comprising:
A difference process is performed between the teaching image and a plurality of captured images of the inspection target object that is known to be good, and for pixels that have exceeded the difference threshold, the difference result is equal to or smaller than the difference threshold. 2. The image processing method according to claim 1, wherein the weighting is performed by calculating a weight in a range of 0 to 1.0. 3. The setting of weight for each pixel of the teaching image is as follows:
The image processing method according to claim 1, wherein the image processing is performed manually. 4. A pattern matching process of the captured image is performed using the teaching image before execution of the difference process, and the position and rotation of the position where the coincidence obtained by the pattern matching process exceeds a predetermined threshold value. 4. The image processing method according to claim 1, wherein a plurality of angles are detected, and the plurality of candidates are set as target portions of the difference processing, and the difference processing is performed on each of the target portions. 5. A method according to claim 1, wherein a pattern matching process of the captured image is performed in advance using the teaching image before the execution of the difference process, and the position and rotation of the position where the degree of coincidence obtained by the pattern matching process exceeds a predetermined threshold value. A plurality of angles are detected, and the above-described difference processing is executed while changing the position and the rotation angle in a region near the position of the plurality of candidates, and the result when the difference area value is minimized for each position of the candidate is a final result. The image processing method according to any one of claims 1 to 3, wherein the image is output as (1). 6. A processing means for performing a difference process for calculating a difference in brightness between each pixel between a captured image of an inspection object captured by a TV camera and a teaching image of the object registered in a memory in advance. In an image processing apparatus that outputs a total area of pixels in which a difference result obtained by calculation exceeds a preset difference threshold, a weight is set for each pixel of the teaching image, and a pixel having a low importance is set. Image processing characterized by comprising means for performing a process for making a difference result equal to a difference threshold, so that an area of the pixel is not counted as an area of a pixel exceeding the difference threshold and is not recognized as a difference. apparatus. 7. The method of setting weights for each pixel of the teaching image,
Difference processing is performed between the teaching image and a plurality of captured images of the inspection object that is known to be non-defective, and for pixels exceeding the difference threshold, weighting is performed so that the difference result is equal to or smaller than the threshold. 7. The image processing apparatus according to claim 6, wherein the calculation is performed by calculating and setting in a range of 0 to 1.0. 8. A method for setting a weight for each pixel of the teaching image,
7. The image processing apparatus according to claim 6, wherein the input setting is performed by manual operation means. 9. Before executing the difference processing, a pattern matching process of the captured image is performed in advance using the teaching image, and a position where the degree of coincidence obtained by the pattern matching process exceeds a predetermined threshold value. 9. A method according to claim 6, further comprising a step of automatically detecting a plurality of candidates for the rotation angle, performing the difference processing on each of the plurality of candidates detected as a target of the difference processing. The image processing device according to any one of the above. 10. A pattern matching process of the captured image is performed using the teaching image before execution of the difference process, and the position and rotation of the position and rotation where the degree of coincidence obtained by the pattern matching process exceeds a predetermined threshold value. A means for automatically detecting a plurality of angles is provided, and the difference processing is executed while changing the position and the rotation angle in a region near the position of the detected plurality of candidates, and the difference area value is minimized for each position of each candidate. 9. The image processing apparatus according to claim 6, further comprising: a unit that outputs a result in the case of a final result.