JPH01245365A - Method and device for evaluating inspection capacity for appearance inspector - Google Patents

Abstract

PURPOSE:To accurately obtain an evaluation value even if a defect having the same density gradation as that of the surface of a substance to be inspected exists by calculating the evaluation value of inspection capacity based on the 1st detection probability, the 2nd appearance probability and distribution probability. CONSTITUTION:The 1st detection probability of an appearance inspector and the 1st appearance probability of a defect on the surface of a substance to be inspected are respectively calculated in each feature value of an image expressing the feature of the image of the substance to be inspected. Then the 2nd inspection probability of the appearance inspector and the 2nd appearance probability of the defect are found out in each degree of the image expressing the feature of the image of the substance to be inspected and the distribution probability in the degree of a feature value other than the degree of the feature value included in the image data of the substance to be inspected is calculated based on the 2nd detection probability and the 2nd appearance probability. The evaluation value of inspection capacity is calculated based on the 1st detection probability, the 1st appearance probability and the distribution probability. Even if a defect having the density gradation as that of the surface of the substance to be inspected exists, the evaluation value can be accurately obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention is directed to an appearance inspection device for photographing an object to be inspected and inspecting defects on the surface of the object based on the data of the photographed image. This invention relates to a testing ability evaluation method and device.

[Prior Art] Conventionally, an object to be inspected is imaged using a two-dimensional camera, or an image is taken using a one-dimensional camera while moving the object to be inspected in a direction perpendicular to the scanning direction of the one-dimensional camera. Various measurement methods, defect detection algorithms, visual inspection equipment, etc. are used to detect defects such as dirt, scratches, damage, and missing parts on the surface of the inspected object from the two-dimensional grayscale or colored images obtained by this method. has been devised.

A method and apparatus for evaluating the inspection ability of these visual inspection apparatuses have been proposed by the present inventor in Japanese Patent Application No. 63-031444. This inspection method for an appearance inspection apparatus is an inspection ability evaluation method for an appearance inspection apparatus that photographs an object to be inspected as a grayscale image and inspects defects on the surface of the object to be inspected based on the data of the grayscale image.
The detection probability, which is the probability that the appearance inspection device can detect an image of each of the above-mentioned gradations, is calculated based on the image data including all the above-mentioned gradations of gradation for each predetermined gradation of gradation, and The appearance probability, which is the probability that the defect will appear, is calculated based on the data of the image of the object to be inspected that has a defect on the surface for each gradation, and the detection probability and the appearance probability are calculated for each gray level. A first summation, which is the sum of the products of the above, is calculated over all the above-mentioned density gradations, and a second summation, which is the sum of the above-mentioned appearance probabilities of each of the above-mentioned density gradations, is calculated over all the above-mentioned density gradations. The first total sum is divided by the second total sum, and the divided value is used as an evaluation value of the inspection ability of the visual inspection apparatus.

[Problems to be Solved by the Invention] However, in the above-mentioned conventional inspection performance evaluation method and apparatus for the visual inspection apparatus, dirt on the surface of the object to be inspected is detected based on the gray scale as described above. Therefore, if the gradation of the dirt attached to the object to be inspected is the same as the gradation of gradation of the gradation of the surface of the object to be inspected, the dirt cannot be detected. There was a problem in that an error occurred in the evaluation value of the inspection device.

An object of the present invention is to solve the above problems, and even if the gradation of the gradation of defects such as dirt attached to the object to be inspected is the same as the gradation of gradation of the gradation of the surface of the object to be inspected, it is possible to An object of the present invention is to provide an evaluation method and device for a visual inspection device that can accurately and automatically quantify and evaluate inspection ability.

"Means for Solving the Problems" The evaluation method of the first invention is for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image. In the inspection ability evaluation method, the first detection is the probability that the appearance inspection device can detect an image of each gray scale based on image data including all the gray scales for each predetermined gray scale. A probability is calculated, and a first appearance probability, which is a probability that the defect appears, is calculated based on data of an image of the object to be inspected having a defect on the surface for each gray level, and the first detection is performed. A first sum, which is the sum of the products of the probability and the first appearance probability, for each of the gray scales, is calculated over all the gray scales, and A second sum, which is the sum of appearance probabilities, is calculated over all the gray scales, the first sum is divided by the second sum, and the result is calculated based on the image data of the object to be inspected. Calculate the probability of appearance of image data having grayscale gradations of the object, and calculate the probability of appearance of image data having gradations of gradation of the object, and calculate the probability of appearance of image data having gradations of gradation of the object, and calculate the probability of appearance of image data that includes all degrees of the feature amount for each degree of the feature amount of the image representing the image characteristics of the object to be inspected. A second detection probability is calculated, which is a probability that the appearance inspection device can detect an image with each degree of the feature amount, and the defect appears for each degree of the feature amount based on the image data of the object to be inspected. A second appearance probability is calculated, which is a probability that The distribution probability is calculated, and the division value is added to the product of the probability of appearance of image data having grayscale gradation of the object to be inspected and the distribution probability, and the added value is calculated as It is characterized by being an evaluation value.

The evaluation device of the second invention is an inspection ability evaluation device for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image. a first calculation means for calculating a first detection probability that is a probability that the appearance inspection apparatus can detect an image of each density gradation based on image data including all of the density gradations for each density gradation; and a second calculation means for calculating a first appearance probability, which is a probability that the defect will appear, based on data of an image of the object to be inspected having a defect on its surface for each density gradation; 1 detection probability and the first appearance probability for each of the gray scales; a fourth calculation means for calculating a second sum which is the sum of the first appearance probabilities of the gray scales over all the gray scales; and dividing the first sum by the second sum. a sixth calculation means for calculating the probability of appearance of image data having gradations of the to-be-inspected object based on the image data of the to-be-inspected object; a second detection which is the probability that the visual inspection device can detect an image with each degree of the feature amount based on image data including all the degrees of the feature amount of the image representing the feature; a seventh calculation means for calculating a probability; and an eighth calculation means for calculating a second appearance probability, which is a probability that the defect will appear for each degree of the feature amount, based on the image data of the object to be inspected. and a ninth calculation means for calculating the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on the second detection probability and the second appearance probability. and a tenth calculation means for adding the division value calculated by the fifth calculation means to the product of the appearance probability of the image data having the density gradation of the object to be inspected and the distribution probability, The present invention is characterized in that the added value obtained by the tenth calculation means is used as an evaluation value of the inspection ability of the visual inspection device.

The evaluation method of the third invention is an inspection ability evaluation method for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the image taken. A first detection probability, which is a probability that the appearance inspection device can detect an image of each density gradation, is calculated based on image data including all of the above-mentioned density gradations for each density gradation, and A first appearance probability, which is the probability that the defect will appear, is calculated based on the image data of the object to be inspected that has a defect on its surface for each key, and the first detection probability and the first appearance probability are calculated. A first sum, which is the sum of the products of each of the gray scales with a predetermined first importance level for each gray scale of the object to be inspected, is calculated over all of the gray scales. and the first appearance probability and the first
The second value, which is the sum of the products of each gray level, with the importance of
is calculated over all the gray scales, the first sum is divided by the second sum, and the gray scale of the object to be inspected is obtained based on the image data of the object to be inspected. The appearance inspection device calculates the appearance probability of the image data, and calculates the feature amount based on the image data including all degrees of the feature amount for each degree of the feature amount of the image representing the feature of the image of the object to be inspected. A second detection probability, which is the probability of detecting an image with each degree of Calculate the probability and calculate the second
The feature amount that the image data of the object to be inspected has based on the detection probability of calculating the distribution probability of the degree of the feature amount other than the degree of , and adding the division value to the product of the probability of appearance of the image data having the gray scale of the object to be inspected and the distribution probability,
It is characterized in that the above-mentioned added value is used as an evaluation value of the inspection ability of the above-mentioned appearance inspection device.

An evaluation device according to a fourth aspect of the invention is an inspection ability evaluation device for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image. a first calculation means for calculating a first detection probability that is a probability that the appearance inspection apparatus can detect an image of each density gradation based on image data including all of the density gradations for each density gradation; and a second calculation means for calculating a first appearance probability, which is a probability that the defect will appear, based on data of an image of the object to be inspected having a defect on its surface for each density gradation; 1 detection probability, the first appearance probability, and a predetermined first importance level for each density gradation for the object to be inspected. a third calculating means for calculating the sum of the values over all the gray levels, and a second calculating means that is the sum of the products of the first appearance probability and the first importance level for each gray level. based on the image data of the object to be inspected; a sixth calculating means for calculating the probability of appearance of image data having gradations of gradations of the object to be inspected; a seventh calculation means for calculating a second detection probability that is a probability that the appearance inspection apparatus can detect an image of each degree of the feature amount based on image data including all the degrees; and an image of the object to be inspected. eighth calculation means for calculating a second appearance probability that is the probability that the defect will appear for each degree of the feature amount based on the data; and the second detection probability, the second appearance probability, and the second appearance probability. distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on a second degree of importance predetermined for each degree of the feature amount for the object to be inspected; and the fifth calculation means for calculating the product of the appearance probability of the image data having the gray scale of the object to be inspected and the distribution probability.
The first O which adds the division value calculated by the calculating means of
, and the added value calculated by the first O calculation means is used as an evaluation value of the inspection ability of the visual inspection apparatus.

The evaluation method of the fifth invention is an inspection ability evaluation method for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image. A first appearance probability, which is the probability that the defect will appear, is calculated based on the data of the image of the object to be inspected that has a defect on its surface for each gray scale, Calculate a first summation, which is the sum of the products of the first detection probability, which is the probability of detecting the gray scale image, and the first appearance probability, for each of the gray scale levels, over all the gray scale levels. Then, a second sum, which is the sum of the first appearance probabilities of each of the gray scales, is calculated over all the gray scales, and the first sum is calculated as the second sum of the first appearance probabilities for each gray scale.
Based on the image data of the object to be inspected, calculate the appearance probability of image data having gradations of gradation of the object to be inspected, A second appearance probability, which is the probability that the defect will appear, is calculated for each degree of the feature amount of the image representing the feature of the image, and the appearance inspection device detects an image with each degree of the feature amount determined in advance. Based on the second detection probability, which is the probability of occurrence, and the second appearance probability, a distribution probability of the degree of the feature other than the degree of the feature that the image data of the object to be inspected is calculated, and The present invention is characterized in that the above-mentioned division value is added to the product of the appearance probability of image data having the gray scale of the object and the above-mentioned distribution probability, and the above-mentioned added value is used as an evaluation value of the inspection ability of the above-mentioned appearance inspection apparatus.

゛The evaluation device of the sixth invention is an inspection ability evaluation device for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image. a first calculation means that calculates a first appearance probability that is a probability that the defect will appear based on data of an image of the object to be inspected that has a defect on its surface for each density gradation; The first sum, which is the sum of the products of the first detection probability, which is the probability that the inspection device can detect an image of each density gradation, and the first appearance probability, for each density gradation, is calculated by The second calculation is performed over the above gray scale.
a third calculation means for calculating a second sum which is the sum of the first appearance probabilities for each of the gray scales over all the gray scales; a fourth calculation means for dividing by the second summation; and a fifth calculation means for calculating the appearance probability of image data having grayscale gradations of the object to be inspected based on the image data of the object to be inspected. , a sixth step of calculating a second appearance probability, which is a probability that the defect appears for each degree of image feature 1 representing the feature of the image of the object to be inspected, based on the image data of the object to be inspected;
image data of the object to be inspected is calculated based on the second detection probability, which is a predetermined probability that the appearance inspection device can detect an image of each degree of the feature amount, and the second appearance probability. seventh calculating means for calculating the distribution probability of the degree of the feature amount other than the degree of the feature amount having the feature amount; and an eighth calculation means for adding the division value calculated by the fourth calculation means, and the added value obtained by the eighth calculation means is used as an evaluation value of the inspection ability of the appearance inspection device. It is characterized by

The evaluation method of the seventh invention is an inspection ability evaluation method for an appearance inspection apparatus that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image. A first appearance probability, which is the probability that the defect will appear, is calculated based on the data of the image of the object to be inspected that has a defect on its surface for each gray scale, a first detection probability that is a probability of detecting a gray scale image; a first appearance probability; and a predetermined first importance level for each gray scale for the object to be inspected; A first sum, which is the sum of the products for each tone, is calculated over all the gray scales, and the sum of the products of the first appearance probability and the first importance level for each gray scale is calculated. A second summation is calculated over all the gray scales, the first summation is divided by the second summation, and the grayscale of the object to be inspected is calculated based on the image data of the object to be inspected. The probability of appearance of image data having gradations is calculated, and the probability that the defect appears for each degree of the feature amount of the image representing the feature of the image of the object to be inspected is calculated based on the image data of the object to be inspected. A second probability of appearance is calculated, and a second detection probability that is a predetermined probability that the appearance inspection device can detect an image with each degree of the feature amount, a second probability of appearance, and a second probability of occurrence of the object to be inspected are calculated. calculate the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on a second degree of importance determined in advance for each degree of the feature amount, and The division value is added to the product of the appearance probability of image data having gradations of gradation of the object to be inspected and the distribution probability, and the added value is used as an evaluation value of the inspection ability of the appearance inspection device. do.

An evaluation device according to an eighth aspect of the present invention is an inspection ability evaluation device for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image. a first calculation means that calculates a first appearance probability that is a probability that the defect will appear based on data of an image of the object to be inspected that has a defect on its surface for each degree gradation;
a first detection probability that is predetermined and is a probability that the appearance inspection device can detect an image of each gray scale, the first appearance probability, and a predetermined probability for each gray scale for the inspected object. a second calculation means for calculating a first sum, which is a sum of the products of the first importance level and each of the gray scale levels, over all the gray scale levels; and the first probability of occurrence and the a third calculating means for calculating a second sum, which is the sum of the products of each gray scale with the first importance level, over all the gray scales; 2, a fifth calculation means for calculating the probability of appearance of image data having a gray scale of the object to be inspected based on the image data of the object to be inspected; a sixth calculation means for calculating a second appearance probability, which is a probability that the defect will appear, for each degree of the feature amount of the image representing the feature of the image of the object to be inspected, based on the image data of the object to be inspected; , a second detection probability that is predetermined and is a probability that the appearance inspection device can detect an image with each degree of the feature amount, the second appearance probability, and each degree of the feature amount for the object to be inspected. a seventh calculating means for calculating the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on a second degree of importance determined in advance; and an eighth calculation means for adding a division value calculated by the fourth calculation means to the product of the appearance probability of image data having gray scale of the object and the distribution probability, and the eighth calculation means It is characterized in that the added value calculated by the means is used as an evaluation value of the inspection ability of the visual inspection device.

Furthermore, each of the above inventions is characterized in that the feature amount of the image is the number of pixels corresponding to the area of a dirt defect on the surface of the object to be inspected, or the circumference length of the object to be inspected.

[Operation] According to the first and second inventions, there is an inspection ability evaluation method for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on data of the photographed image. In this step, a first detection probability, which is a probability that the appearance inspection device can detect an image of each density gradation, is calculated based on image data including all of the density gradations for each predetermined density gradation. , calculate a first appearance probability, which is the probability that the defect will appear, based on the data of the image of the object to be inspected having a defect on the surface for each density gradation, and calculate the first appearance probability, which is the probability that the defect will appear, and calculate the first probability of occurrence and the first detection probability. The first sum, which is the sum of the products of each gray scale with the appearance probability of 1, is calculated by adding all the 77! A second sum, which is the sum of the first appearance probabilities of each of the lightness and lightness levels, is calculated over all the lightness and lightness levels, and the first sum is calculated as the second summation, which is the sum of the first appearance probabilities for each of the lightness and lightness levels. and calculate the appearance probability of image data having gray scale of the object to be inspected based on the image data of the object to be inspected;
Probability that the visual inspection device can detect an image with each degree of the feature amount based on image data including all degrees of the feature amount for each degree of the feature amount of the image representing the feature of the image of the inspected object. A second detection probability is calculated, which is the probability that the defect appears for each degree of the feature amount, based on the image data of the inspection object, and a second detection probability is calculated, which is the probability that the defect appears for each degree of the feature amount, Calculating the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the inspection object has based on the detection probability and the second appearance probability,
The above-mentioned division value is added to the product of the appearance probability of the image data having the grayscale gradation of the above-mentioned object to be inspected and the above-mentioned distribution probability, and the above-mentioned added value is used as an evaluation value of the inspection ability of the above-mentioned appearance inspection apparatus. As a result, the evaluation value for evaluating the inspection ability of the above-mentioned appearance inspection apparatus can be adjusted to can be obtained accurately and quickly.

Further, according to the third and fourth inventions, an inspection ability evaluation method for an appearance inspection apparatus that photographs an object to be inspected and inspects defects on the surface of the object based on data of the photographed image, A first detection probability, which is a probability that the appearance inspection device can detect an image of each density gradation, is calculated based on image data including all of the density gradations for each predetermined density gradation, and A first appearance probability, which is the probability that the defect will appear, is calculated based on the image data of the object to be inspected that has a defect on its surface for each density gradation, and the first probability of occurrence and the first probability of occurrence of the defect are calculated. The first sum, which is the sum of the products of the appearance probability and the predetermined first importance level for each density gradation for the object to be inspected, for each density gradation is calculated for all the density gradations. A second sum, which is the sum of the products of the first appearance probability and the first importance level for each of the gray levels, is calculated over all the gray levels, and the The first sum is divided by the second sum, and the probability of appearance of image data having the density gradation of the object to be inspected is calculated based on the image data of the object to be inspected. a second detection which is the probability that the appearance inspection device can detect an image with each degree of the feature amount based on image data including all degrees of the feature amount for each degree of the feature amount of the image representing the feature; A probability is calculated, and a second appearance probability, which is the probability that the defect appears for each degree of the feature amount, is calculated based on the image data of the object to be inspected, and the second appearance probability is calculated based on the image data of the object to be inspected. The probability of appearance of and the predetermined second
Calculate the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on the importance of The above-mentioned division value is added to the product of and the above-mentioned distribution probability, and the above-mentioned added value is used as an evaluation value of the inspection ability of the above-mentioned appearance inspection apparatus. As a result, the evaluation value for evaluating the inspection ability of the above-mentioned appearance inspection device can be determined by determining whether the gray scale of defects such as dirt attached to the inspected object is the same as the gray scale of the surface of the inspected object. can also be obtained accurately and quickly.

Furthermore, according to the fifth and sixth inventions, in an inspection ability evaluation method for an appearance inspection apparatus that photographs an object to be inspected and inspects defects on the surface of the object based on data of the photographed image. , a first appearance probability, which is the probability that the defect will appear, is calculated based on the data of the image of the object to be inspected having a defect on the surface for each predetermined gradation level, and the predetermined appearance inspection apparatus The first sum, which is the sum of the products of the first detection probability, which is the probability of detecting an image of each density gradation, and the first appearance probability, for each density gradation, is calculated for all the gradations. A second sum, which is the sum of the first appearance probabilities of each of the gray levels, is calculated over all of the gray levels, and the first sum is calculated as the second total. The appearance probability of the image data having the gray scale of the object to be inspected is calculated based on the image data of the object to be inspected. A second appearance probability, which is the probability that the defect will appear, is calculated for each degree of the feature amount of the image representing the feature of the image, and is determined in advance so that the appearance inspection device can detect images with each degree of the feature amount. Based on the second detection probability, which is a probability, and the second appearance probability, a distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has is calculated, and The above-mentioned division value is added to the product of the appearance probability of the image data having the density gradation and the above-mentioned distribution probability, and the above-mentioned added value is used as an evaluation value of the inspection ability of the above-mentioned visual inspection apparatus. As a result, even if the evaluation value for evaluating the inspection ability of the above-mentioned appearance inspection device is the same as the gradation level of defects such as dirt attached to the object to be inspected or the gradation level of the gradation level of the surface of the object to be inspected, can be obtained accurately and quickly.

Furthermore, according to the seventh and eighth inventions, an inspection ability evaluation method for an appearance inspection apparatus that photographs an object to be inspected and inspects defects on the surface of the object based on data of the photographed image. , a first appearance probability, which is a probability that the defect will appear, is calculated based on data of an image of the object to be inspected having a defect on the surface for each predetermined gray level, and a first appearance probability is determined in advance, and the first appearance probability is determined in advance by the appearance inspection apparatus. is the probability that an image of each gray level can be detected; the first appearance probability; and a predetermined first importance level for each gray level for the object to be inspected. A first sum, which is the sum of the products for each of the gray scales, is calculated over all the gray scales, and the first summation of the first appearance probability and the first importance level is calculated for each gray scale. A second sum, which is a sum of products for each, is calculated over all the gray levels, and the first sum is divided by the second sum, and the result is calculated based on the image data of the test object. The appearance probability of image data having gradations of gradation of the inspection object is calculated, and the defect is calculated for each degree of the image feature amount representing the image feature of the inspection object based on the image data of the inspection object. A second appearance probability, which is a probability of appearance, is calculated, and a second detection probability, which is a predetermined probability that the appearance inspection device can detect an image with each degree of the feature amount, the second appearance probability, and the above. distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on a second degree of importance predetermined for each degree of the feature amount for the object to be inspected; , and add the division value to the product of the probability of appearance of the image data having the density gradation of the object to be inspected and the distribution probability, and use the added value as the evaluation value of the inspection ability of the appearance inspection device. do. As a result, the evaluation value for evaluating the inspection ability of the above-mentioned appearance inspection apparatus can be adjusted to can be obtained accurately and quickly.

[Example] Fig. 1 shows a visual inspection function and an evaluation device (hereinafter referred to as
It is called appearance inspection and evaluation equipment. ) is a block diagram of.

In FIG. 1, this visual inspection and evaluation apparatus includes a light projecting device 20 that irradiates an object to be inspected with illumination light, a video camera 11, and an analog-to-digital converter (hereinafter referred to as A/D converter).
It's called conversion. ) An imaging device 100 having a circuit 12, a central processing unit 1, and a read-only memory (hereinafter referred to as RO
It's called M. ) 2. Random access memory (hereinafter referred to as RA)
It's called M. ) 3, keyboard 4 and display device 5
The image processing device 200 has an inspection mode for inspecting dirt defects on the surface of an object to be inspected using the imaging device 100, and an inspection mode for inspecting dirt defects on the imaging device 100 and other imaging devices. It is characterized by having first and second evaluation modes for evaluating inspection ability.

In FIG. 1, the video camera 11 photographs the inspected object 10 irradiated with light from the light projector 20, and then includes information on a two-dimensional grayscale image of one field (one screen). A black and white television signal is output to the A/D conversion circuit 12. The A/D conversion circuit 12 performs A/D conversion on the manually inputted television signal image in 16 gradations, and outputs the converted image data to the CPUI via the image processing unit 30 of the image processing device 200. Here, the image data output from the A/D conversion circuit 12 has a gradation in which each pixel of one screen is expressed in 16 gradations (hereinafter referred to as gradation, and the gradation is the lightest gradation). The image data of one screen consists of image data of 1000 pixels.

The CPU 1 is a control circuit that controls operations within the image processing device 200, and the ROM 2, RAM 3, and display device 5 are connected to the CPU 1.

The ROM 2 stores a system program for controlling the inspection mode and the first and second evaluation modes in the image processing device 200, and the RAM 3 stores a system program for controlling the inspection mode and the first and second evaluation modes in the image processing device 200.
Used as a work area for the PUI and as a CPU
The image data input to I is temporarily stored. The display device 5 displays each image of the image data manually input to the CPUI and the image data temporarily stored in the RAM 3, and also displays the data of the inspection results and evaluation results calculated by the CPU 1.

The operation of the visual inspection and evaluation apparatus configured as described above will be explained separately in the inspection mode, the first evaluation mode, and the second evaluation mode.

(1) Inspection mode First, the object to be inspected, which has no dirt or defects on its surface, is
1, and the image data of the photographed image (hereinafter referred to as image data of a normal image) is transferred to the A/D conversion circuit 1.
2 and CPIJ l, the data is stored in the RAM 3 and displayed on the display device 5. The image data of the above-mentioned normal image is composed of gray scale data for each pixel for one screen. Then, the CPUI executes the above R
It is detected whether the image data of the normal image stored in AM3 includes pixels of each of the above-mentioned gray scales, and a total of 16 pieces of data (hereinafter referred to as (referred to as gradation data of a normal image) is stored in the RAM 3.

Next, when inspecting for defects on the surface of the object to be inspected, the object to be inspected is photographed using a video camera l, and the image data of the photographed image (hereinafter referred to as image data of the inspection image) is A/D converted. RA via circuit 12 and CPUI
It is stored in M3 and displayed on display device 5. Next, the CPU 1 detects whether or not the image data of the inspection image stored in the RAM 3 includes pixels of each density gradation, and then detects a total of 16 pixels corresponding to each detected density gradation. data (hereinafter referred to as the gradation data of the inspection image) and the gradation data of the normal image stored in the RAM 3, and the gradation data of the inspection image that does not exist in the gradation data of the normal image is detected. Determine whether it is included.

Here, if the gradation data of the inspection image includes a gradation that is not included in the gradation data of the normal image, the CPUI sends the data of the gradation gradation that is not included in the gradation data of the normal image to the display device 5. On the other hand, if the gradation data of the inspection image does not include gradation levels that are not present in the gradation data of the normal image, the CPU determines that there is no dirt defect and displays it as a "stain defect". None'' and display device 5
output and display it.

Furthermore, in the gradation data of the inspection image, the CPUI uses the image data of the normal image and the inspection image for each gradation gradation (hereinafter referred to as a specific gradation gradation) that is not in the gradation data of the normal image. Count the number of pixels in each image data.

Counting the number of pixels for this specific gray level is performed by counting pixels having the specific gray level, for example, into white image data for the image data of the normal image and the image data of the inspection image stored in the RAM 3. On the other hand, by converting pixels other than the specific gray scale to black image data, and counting the white pixels for the image data of the normal image and the image data of the inspection image. can get. This procedure is known in conventional image processing technology as "binarization of image data."

A pixel in which the number of pixels in the image data of the inspection image for the specific gradation level counted is predetermined around the number of pixels in the image data of the normal image for the specific gradation gradation counted above. It is determined by the CPUI whether or not the number is within the allowable range. If each specific density gradation is within the above tolerance range, it is assumed that there is no dirt on the inspection object, and cput is "no dirt".
is displayed on the display device 5. On the other hand, if any of the specific gray scales is not within the allowable range, the CPU will display the data of the specific gray scale that is not within the allowable range as "stained" to the display device 5.
to be displayed.

In the above inspection mode, each specific gray scale of the image data of the test image that is not included in the image data of the test image is performed. The process may also be performed for gradations.

Using the above-mentioned visual inspection and evaluation device using images with 16 gradations of gradation, dirt adhering to the surface of the object to be detected, which has a red company emblem printed on a flat plate with a light yellow background, is inspected. An experimental example in which a visual inspection device was evaluated will be described below. Here, it is assumed that the above-mentioned stains include stains ranging from pale gray to black.

In a normal image taken using the video camera 11 of an object to be inspected with no dirt on its surface, the light yellow ground color is the brightest, resulting in a light gray image, and the company emblem is a slightly darker gray color. For example, the gradation of the ground color is 4.
~6, the gray scale of the mark is 10~12.

In contrast, images of objects to be inspected with dirty surfaces (hereinafter referred to as
It is called a dirty image. ), and for example, if ink is attached to the surface, the gradation level of the nearly pure black image will be, for example, 15. Conversely, when white powder adheres to the surface of the object to be inspected, the gray scale of the image becomes 2. In these cases, it is relatively easy to detect dirt using the above-mentioned inspection device. That is, in normal images, 4-6 and 10-1
However, in addition to the gradation of the normal image, the soiled image has a gradation of 15, which is a gray stain image, and a gradation of 2, which is an image of white powder. It also includes the tone. Therefore, as mentioned above, whether it is a normal image or a dirty image,
It is checked whether the image has only gray scales of 4 to 6 and 10 to 12, or whether it contains other gray scales, and if it does, it can be detected as a dirty image. can.

FIG. 2 shows data showing the relationship between the detection probability of the apparatus and the appearance probability of dirt defects with respect to the density gradation in this case.

Here, the detection probability is a characteristic unique to the inspection device, and the detection probability at each gray level included in a normal image is 0, while the detection probability at each gray level when a dirty image can be reliably detected. The detection probability at each gradation is 1. if,
If the inspection device is not perfect and does not detect an image of the gray scale with a certain probability, the detection probability is less than 1. In the above visual inspection and evaluation device, the detection probability depends on the performance of the imaging device 100.

Further, the appearance probability of a stain defect is the probability that a stain defect at each density gradation will occur among all stain defects that occur in a large number of inspected objects, for example, 100 or more manufactured by the same process. Note that the sum of the above-mentioned appearance probabilities of each gray level is l.

As mentioned above, the normal image includes images with gradations of 4 to 6 and 10 to 12, so as shown in (A) in Figure 2, the images with gradations of 4 to 6 and 10 to 12 are included. The above-mentioned inspection apparatus cannot detect dirt defects in grayscale images.

Furthermore, when the image of the stain defect has only 2 to 15 gray scales, as shown in FIG. Can detect 100% of the test object (hereinafter referred to as 100% detection capability of the test device)
It is said that it is 0%. ) can be expected.

Now suppose that a smudge defect is caused by the red printing ink used for the mark spilling onto the surface of the board. In this case, as shown in FIG. 3(B), the stain image is an image with the same 11 gradations as the mark, so the stain defect cannot be detected by the discrimination based on the above-mentioned density gradation. That is, the above-mentioned inspection device cannot detect the above-mentioned dirt defect. Therefore, the inspection ability of the inspection device is 0%.

Furthermore, as a case of boat fishing, a case will be described in which the dirt defect includes gradations other than the above two examples, as shown in FIG. 4(B). In this case, which gradation of the stain defect from 1 to 16 appears in the image of the tq defect becomes a statistical probability phenomenon. Here, as long as the above-mentioned conventional inspection device based on gradation discrimination is used, the background color and mark of the company emblem are the same as 4.
Images with gray scales of ~6 and 10~12 cannot be detected as stain defects.

Therefore, (i) is expressed by the following equation for the first inspection ability of the inspection system including the inspection device and the object to be inspected.

1:1 f +”F (i)・G(i)di −fl・・F・(i)・G (i)di ”” 1b
) Here, F(i) is the detection probability of the inspection device at the gray scale of j, and G(i) is the probability of appearance of the stain defect on the inspection device at the gray scale of i. Also,
F r (i) is the detection probability of the ideal inspection device at the gray level of i, and is ■.

In other words, the first inspection capability of the inspection system including the inspection device and the object to be inspected is, as shown in equation (1a), the detection probability of the inspection device at each gray scale and the detection probability at that gray scale. The detection probability is 1 for all gradation levels and the value obtained by summing the product value at each gradation level over all gradation levels. In an ideal inspection device, the value obtained by calculating the above-mentioned sum and the value of the ratio are used. Here, if the number of gradations to be investigated is infinitely increased, the detection probability of the inspection device at each gradation is calculated by setting the inspection performance of the inspection system to 1, as shown in equation (1b). Take the product of the appearance probability of a dirt defect at that density gradation,
A value obtained by integrating the product value at each gray scale over all gray scales, and a value obtained by calculating the integral value using the ideal inspection apparatus are used as a ratio value.

If we use the index for the above inspection performance, for example, if the performance of the inspection device is good and the gradation of dirt on the object to be inspected does not overlap with the background color of the company emblem or the gradation of gradation of the mark, then it will be 1. The values will be close to each other, but if the detection probability of the inspection device itself at each gradation level is smaller than 1, or if the probability that the gradation level of a dirt defect overlaps with the gradation level of the background color or mark increases, then The value of 1 for the inspection ability becomes small and close to 0. That is, by using the index of this inspection ability as a guideline, it is possible to indicate whether the inspection system essentially has the inspection ability of the inspection object to be inspected.

Next, in the above-mentioned first inspection performance index of 1, it is assumed that the influence of a stain defect on a product to be inspected is a constant value of 1, regardless of the concentration of the stain defect.

However, in real cases, for example, the degree of deterioration of the image of the product when there is whitish dirt is often evaluated as being smaller than the degree of deterioration of the image of the product when there is black dirt. Therefore, an evaluation that also takes this factor into account is more appropriate.

Therefore, *(i) is expressed as the second inspection ability of the inspection system including the inspection device and the object to be inspected by the following equation.

1, H(i) is a numerical value (hereinafter referred to as the degree of importance) of the degree of deterioration of the image of the product to be inspected due to the stain defect of the gray scale of i; The larger the value of the degree, the greater the image deterioration caused by dirt defects on the inspected object, and the higher the importance of the gradation.

In other words, the second inspection capability of the inspection system including the inspection device and the test object is as shown in equations (2a) and (2b), where each numerator and denominator in equations (la) and (1b) are is multiplied by the above importance level. Here, the degree of importance H(i) is determined in advance based on the value and performance of each object to be inspected, the sense of value of the group handling the object, the manufacturing process, and the like.

Therefore, by using the index of this inspection ability as a guide, we can determine how much inspection ability the inspection system essentially has for the inspection object as a product that gives the inspection object. It is possible to display the image in consideration of the degree of importance indicating the degree of deterioration of the image.

Furthermore, the above equation (1a) can be expressed as the sum of partial sequences divided by gradation, for example, as in the following equation.

1=1 Note that the above equation (1b) can be expressed similarly to equation (3) as a sum of partial integrals divided by gradation.

Table 1 is a table showing the detection probability F (i) and the appearance probability G (i) of a certain inspection device. When (i) is calculated based on equations (2a) and (3) for the first testing ability of this testing device, it becomes K, (i) -0,81.

Here, the detection probability F for gray scales i=5 and 11
(i) −〇, so the second on the right side of equation (3) above is
term and the fourth term are O. That is, as stated in the section of the problem to be solved by the invention, when stains having the same gradation of gradation as the gradation of gradation of the surface of the object to be inspected adhere to the surface of the object to be inspected, The detection probability F (i) is unconditionally O, and the first and second
The inspection performance of , , , does not include evaluation at the above-mentioned density gradation. In this example, below, for the stain having the same gradation 1=11 as the company emblem mark in the above experimental example among the gradation gradation i=5 and 11,
An evaluation regarding the number of pixels at a specific gray scale (in this experimental example, gray scale 1=11) described in the above inspection mode is added. As is well known, this number of pixels corresponds to the area of the pixel at the above-mentioned specific gray level.

At this time, including some measurement errors, the number of pixels corresponding to the area of the mark is always 401 to 500,
It is assumed that an image with a number of pixels outside of this range is an image other than a mark, that is, an image of dirt.

Figure 5 is obtained by calculating the probability that dirt will appear in each determined range of the number of pixels of dirt that appears in an image with the same gradation level of 1 as the mark. Next, the detection probability of the above-mentioned inspection apparatus for an image of 11 gray scales is almost uniform, and even if the number of pixels changes, it can be detected reliably and at a constant value. In other words, if the detection probability is 1 as shown in FIG.
In an inspection device that inspects dirt defects using only grayscale gradations, even if it is an undetectable dirt image with grayscale 11, it is not possible to detect it even if it is an image with an area other than 401 to 500 pixels. If so, it can be detected as a stain defect. Therefore, the inspection ability Ka (i.u, j), which is a function of the number of pixels j at the number of pixels other than the mark at the specific gray level 11, is defined as follows.

...(4) Here, P(j) is the detection probability of the inspection device at the number of pixels j at the gradation level 11, and Q<j) is the detection probability of the inspection device at the number of pixels j at the gradation level 11. Pr (j) is the probability of occurrence of the above number of pixels, and Pr (j) is the detection probability of an ideal inspection device for the number J of pixels at the gray level 11, and has a value of 1.

That is, as shown in equation (4) for the inspection ability K a (i = 11. The value obtained by summing the product value for each pixel number j over the number of pixels other than the number of pixels of the mark above, and the ideal detection probability where the detection probability is 1 for all pixel numbers j. The value of the ratio of the above sum to the value obtained using a suitable inspection device is used. Here, if the number j of pixels to be investigated is infinitely increased, equation (4) expressed by the above summation equation can be expressed by an integral equation.

Furthermore, the inspection ability Ka expressed by the above equation (4) is multiplied by the following equation, which is the appearance probability S of the appearance of a grayscale image with gradation level 11: +=1 Furthermore, the above multiplication result Ka-3 is By adding 1 to the above-mentioned first inspection ability, 3 is expressed as the third inspection ability indicating the evaluation of the inspection ability of the above-mentioned inspection device using the following formula.

... (6) Furthermore, similar to the above inspection ability Kt, the importance level H(i), which quantifies the degree of deterioration of the image of the product caused by the dirt defect of the density gradation i on the inspected object, and the specific Considering the importance R(j) as a function of the number of pixels j in the gray scale of 1
4 is obtained by multiplying the denominator and numerator of the first term on the right side of the above equation (6) by the importance level H(i) which is a function of the density gradation i, and the above (6). ), the importance R(j
) to obtain the following formula.

... (7) In addition, in the first evaluation mode shown below, evaluation was performed using 3 for the third inspection ability, and also in the second evaluation mode shown below.
In the evaluation mode, evaluation is performed using 4 for the fourth inspection ability.

In addition, the degree of importance R(j), which depends on the number of pixels j, is determined in advance based on the value and performance of each object to be inspected, the sense of value of the group handling the object, the manufacturing process, etc., similar to the degree of importance H(i) above. It is determined.

Therefore, 3. By using index 4 as a guideline, you can determine whether or not the inspection system essentially has the ability to inspect the object to be inspected. Even if the same stain defect as the sample exists on the surface of the object to be inspected, it can be properly evaluated.

For example, 3. The closer , to 1 is, the more complete the inspection system is.However, in a normal inspection system, if , , and , are, for example, 0.4 or more, the inspection system has a valuable inspection ability. It is considered a system. In addition, 3. For example, 4 is 0
．． If it is 6 or more, it is a highly valuable inspection system, for example, if it is 0°8 or more, it can be said that it is an extremely valuable inspection system.

(2) First evaluation mode First, in order to obtain the detection probability F (i) of the imaging device 100, an image having pixels of all 16 gray scales (hereinafter referred to as a reference image) is A photograph is taken using the camera 11, and image data of a reference image, which is image data of the photographed image, is stored in the RAM 3 via the A/D conversion circuit 12 and the CPUI. Next, the CPU detects whether or not the image data of the reference image stored in the RAM 3 includes pixels of each gray scale, and detects the detected gray scale of the reference image, which is a total of 16 pieces of data. data to RAM
Store in 3.

Next, the procedure for obtaining the gradation data of the reference image described above is repeated, for example, 10 times or more, and the average value of the gradation data of the reference image at each of the gradation levels is determined by the imaging device 10.
It is stored in the RAM 3 as a detection probability F(i) of 0.

Furthermore, an object to be inspected with no dirt or defects on its surface is photographed in advance using the video camera II, and gradation data of a normal image is determined in the same manner as in the above inspection mode and stored in the RAM 3.

Next, the object to be inspected with the stain defect is photographed using the video camera 11, and the image data of the photographed image (hereinafter referred to as image data of the stain image) is stored in the RAM 3 via the A/D conversion circuit 12 and the CPUI. is stored in Then, C
After detecting whether or not the image data of the dirt image stored in the RAM 3 includes pixels of each gray scale, the PUI detects a total of 16 pieces of detected data (hereinafter referred to as the gray scale of the dirt image). ) is compared with the gradation data of the normal image stored in the RAM 3, and it is determined whether the gradation data of the dirty image includes gradations that are not included in the gradation data of the normal image. . Here, if the gradation data of the dirty image includes a gradation level that is not included in the gradation data of the normal image, cpul sets the dirt appearance data at the gradation level to 1; When the gradation data of an image does not include a gradation level that is not present in the gradation data of a normal image, the CPUI sets the dirt appearance data at the gradation level to O, and calculates the data for all gradation levels. A total of 16 pieces of dirt appearance data are stored in the RAM3.

Next, after repeating the above-described procedure for obtaining stain appearance data on, for example, 100 or more test objects manufactured in the same manufacturing process and having stain defects on the surface,
The average value of the stain appearance data in each of the gray scales is determined, and the average value is stored in the RAM 3 as the appearance probability G(i) of stain defects at each gray scale for the object to be inspected.

Furthermore, the CPUI calculates the detection probability F (i) of the imaging device 100 stored in the RAM 3 and the appearance probability G of the dirt defect.
(i) is read out, and the first testability is calculated using the above equation (1a). As a result, the value of the first term on the right side of the above equation (6) can be determined.

Furthermore, within the field of view of the video camera 11, for example, one
Inspection objects having company emblems having a predetermined number of eight stain defects of 0aI or more are arranged, and the N inspection objects are photographed using the video camera 11. Image data of the photographed image (hereinafter referred to as image data of inspection image N) is stored in RAM 3 via A/D conversion circuit 12 and CPU 1. The above procedure is repeated, for example, an integer number M times of 10 or more.

Next, the CPUI reads the M stored in the RAM3.
The number of pixels at each gray scale i is counted based on the image data of the test images N. Table 3 shows the number of pixels counted above. Table 3 shows the distribution of the number of pixels at each gray scale i, and also shows the distribution of the number j of pixels at gray scale 11, which is the same as the mark, in the column for gray scale 11 in Table 3. It can be obtained from the data on the number of pixels in . That is, by extracting the data regarding the number of pixels j at gradation level 11 from Table 3, Table 5 can be created. Table 5 shows the distribution of image data divided into 5 pixel numbers, and thereby each pixel number j in the product of the second term of the above equation (7)
The appearance probability Q(j) of the image data can be determined for the image data.

Furthermore, based on the image data of the reference image stored in the RAM 3 and the image data of the inspection image N, the CPU The detection probability P(j) at each pixel number j is determined by comparing the distribution at each pixel number j with the distribution at each pixel number j in the image data of the inspection image N.

In Table 3 above, the gradation level i of the company emblem mark
From each data of the same gray scale, for example 11, the number of pixels AO corresponding to the size of one mark is calculated.
By subtracting the product of N and the number of marks, a modified data table is obtained in Table 4.

That is, by this subtraction, Table 4 is constructed from data only on dirt defects other than the marks in the company emblem. From the data in Table 4, for example, the gray level i is 11
By finding the value of the ratio of the number of pixels of image data at and the number of pixels of image data at all gray scales,
The probability of appearance of the image data of the dirt defect at the gradation level 11, which is the value of the term on the left side of the product on the right side of the above equation (6), can be determined.

Therefore, by substituting each value obtained above into the above equation (6), it is possible to obtain 3 as the third inspection ability. The CPUI outputs the data of 3 to the third testability calculated above to the display device 5 for display. This completes the operation of the first evaluation mode.

In addition, for example, 3 is 0. If it is 4 or more, it means that the inspection ability of the inspection system is above the specified level, and it is considered to have passed.On the other hand, if the inspection ability is 3, it is 0. If it is less than 4, the inspection ability of the inspection system is considered to be less than a predetermined value and the inspection system is rejected. Before calculating 3, the evaluation criteria of the inspection system is entered in advance from the keyboard 4 to the above inspection ability, and then stored in the RAM 3 via cput. However, after calculating the inspection performance of 3, the CPU 1 may determine whether the test is pass or fail according to the evaluation criteria, and the evaluation result may be output to the display device 5 for display. Here, the evaluation standard of 3 for inspection performance is O14, but it is not limited to this, but includes inspection process, manufacturing process, degree of combined use with visual inspection, inspection cost and reliability, equipment cost of automatic visual inspection equipment, etc. It may be changed with consideration.

In the above-described operation in the first evaluation mode, the imaging device 1
Although 3 is calculated as the third inspection ability of 00, the above detection probability changes by changing the imaging device, and the above appearance probability changes by changing the inspected object. , another imaging device equipped with a video camera and an A/D conversion circuit is connected to the connector 30 of the image processing device 200, and 3 is calculated as the third inspection capability for the same or another object to be inspected. Good too.

(3) Second evaluation mode At this time, as in the first evaluation mode, the CPUI calculates the detection probability F(i) of the imaging device 100 and the inspection target that is manufactured in the same manufacturing process and has a dirt defect. The appearance probability G(i) of a stain defect is calculated and stored in the RAM 3.

Next, the operator determines the degree of importance H(i) at each gray level i, which indicates the degree of deterioration of the product, and the degree of importance H(i) at each pixel number J, which is predetermined based on the above-mentioned criteria for the object to be inspected. Input the importance level R(j) using the keyboard 4,
The data with the above importance levels H(i) and R(j) are stored in the CPU l
The data is stored in the RAM 3 via the .

Furthermore, as in the first evaluation mode described above, the CPUI calculates the detection probability P(j) and the appearance probability Q for each number of pixels j.
In addition to finding (j), for example, the second value on the right side of the above equation
After determining the value of the term on the left side of the product of the terms, each data is stored in the RAM 3.

Finally, the CPU converts the detection probability F (i), P (j) of the imaging device 100 stored in the RAM 3 into the appearance probability G (i), Q (j) of the stain defect and the above importance level H (i). .

After reading out R(j) and the appearance probability of dirt defects at the above-mentioned density gradation 11 and calculating 4 as the fourth inspection ability using the above equation (7), 4 is added to the fourth inspection ability. The data is output to the display device 5 and displayed. That's it for the second
The evaluation mode operation ends.

In addition, when the second inspection ability of the inspection apparatus of the experimental example in Table 1 is calculated using the above inspection apparatus and evaluation apparatus, it is 0.81.

Table 2 is a graph showing the detection probability, appearance probability, and importance level in an experimental example in which the above-mentioned company emblem was evaluated using this inspection and evaluation device.
The appearance probability and importance are shown in FIG. 6, respectively.

When 1 to 4 are calculated for the first to fourth inspection abilities of the inspection apparatus of the experimental example in Table 2 using the above inspection apparatus and evaluation apparatus, the results are 0. 81. 0. 81.0.90
49,0.9321. Therefore, 3. By finding 4 in , it is possible to obtain the inspection ability that allows evaluation of the inspection ability in consideration of dirt defects having the same gray scale as the normal image.

As explained above, in a visual inspection and evaluation device that inspects contamination defects on an object to be inspected and evaluates its inspection ability, it is possible to automatically calculate 3°K4 for the third or fourth inspection ability. Therefore, as described above, it is possible to obtain an evaluation value of the inspection performance that takes into account the dirt defects that have the same gray scale as the normal image, without or in addition to taking into account the above-mentioned importance level, and to perform the above-mentioned appearance inspection. It is possible to quickly and accurately evaluate the performance of the device. Therefore, when introducing this type of inspection device, it becomes easy to judge which inspection ability is best to introduce.

Furthermore, as a result, it can be expected that the above-mentioned transactional troubles will be reduced, which is extremely meaningful from an industrial perspective.

In the above embodiments, equations (6) and (7) describe the case where the gradation of the gradation of the mark on the company emblem is the same as the gradation of gradation of dirt on the surface of the company emblem. However, the above equations (6) and (7) are not limited to this, but when the gradation of the gradation of the surface of the object to be inspected is the same as the gradation of the gradation of dirt attached to the surface of the object to be inspected. Easily expandable.

In the above embodiments, an apparatus is described that uses a grayscale image of the object to be inspected to check whether there is a dirt defect on the surface of the object. It may be possible to inspect flaws, the normality of the shape of the component, the normality of the position, direction, and orientation of the component, the size of the object to be inspected, the length of the object to be inspected, and the like. Therefore, instead of the number j of pixels of a dirt defect, a continuous or discrete parameter (hereinafter referred to as image feature amount ) may be used.

The feature values of this image include the perimeter of the object, the ratio of the perimeter to the area of the object, the number of pixels at each gray level, the number of holes in the object, the perimeter of the hole, the area of the hole, and the object. circularity count of the object, shape count of the object, Fuere diameter of the object, first moment of the object, second moment of the object, moment relative to the area of the object, coordinates of the center of gravity of the object, basic statistical values of the object, object's relative to the entire image. Occupancy rate, angle between the maximum diameter of the object and a specific axis, maximum brightness of the object, minimum brightness of the object, diameter of the object, volume of the object, axial length of an elliptical object, axial length ratio of an elliptical object, two axes of the object average diameter of 2 in the object
The edge distance between points, the distance between two points on an object, the three-point angle on the object, the four-point angle on the object, the number of multiple objects, the distance between multiple objects, etc. may be used.

In the above embodiment, the video camera 11 is used, but the present invention is not limited to this, and other types of cameras capable of capturing a single screen of grayscale images may be used. Furthermore, a color video camera may be used as the video camera 11, and a colored image containing color information may be used for evaluation instead of a grayscale image.

In the above embodiments, the detection probability of the imaging device 100 is calculated in the first and second evaluation modes of the visual inspection and evaluation device, but the detection probability is not limited to this, and the detection probability calculated by another device or The third and fourth inspection abilities may be calculated using a predetermined detection probability.

In the above embodiment, the A/D conversion circuit 12 A/D converts the image of the television signal manually inputted from the video camera 11 in 16 gradations, but the A/D conversion circuit 12 is not limited to this.
A/D conversion may be performed at a plurality of gradations such as 64, 128, 256, etc. At this time, the data processed by the image processing device 200 is processed at each of a plurality of gray levels.

Table 1 Table 2 Table 3 Table 4 Table 5 [Effects of the Invention] As detailed above, according to the present invention, the first detection probability of the visual inspection device and the detection of surface defects on the inspected object After calculating and determining the first appearance probability for each of the image characteristics of the object to be inspected and the characteristic amount of the image representing the image, the appearance inspection apparatus A second detection probability and a second appearance probability of the defect are determined, and based on the second detection probability and the second appearance probability, the inspection object is determined to have a degree other than the degree of the feature amount that the image data has. The distribution probability of the degree of the feature amount is calculated, and the evaluation value of the inspection ability of the inspection device is calculated based on the first detection probability, the first appearance probability, and the distribution probability. To accurately and quickly obtain an evaluation value for evaluating the inspection ability of the above-mentioned appearance inspection device even if there is a defect such as dirt having the same gradation of gradation as that of the surface of the object to be inspected. Can be done.

In addition to the first detection probability, the first appearance probability, and the distribution probability, a first importance level predetermined for each gray scale and a first importance level predetermined for each feature quantity are also provided. second
Since the evaluation value of the inspection ability of the above-mentioned inspection device is calculated based on the importance of can be obtained accurately and quickly.

Furthermore, the same effect can be obtained when the evaluation value of the inspection ability of the inspection apparatus is calculated using predetermined values of the first and second detection probabilities.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a visual inspection and evaluation device that is an embodiment of the present invention, and FIGS. 2 to 6 are graphs showing evaluation results in experimental examples evaluated using the above evaluation device, respectively. 1...Central processing unit (CPU), 2...Read only memory (ROM), 3...Random access memory (RAM), 4...Keyboard, 5...Display device, IO・...Object to be inspected, 11...Video camera, 12...Analog-to-digital conversion circuit (A/D conversion circuit), 100...Imaging device, 200... Image processing device. Patent applicant Kanebuchi Kagaku Kogyo Co., Ltd. Agent Patent attorney Aohaku Ao Haga 1 person Figure 1 Figure 2

Claims (10)

[Claims] (1) In an inspection capability evaluation method for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, all A first detection probability, which is a probability that the appearance inspection device can detect an image of each density gradation, is calculated based on the image data including the density gradation, and detects a defect on the surface for each density gradation. A first appearance probability, which is the probability that the defect will appear, is calculated based on data of an image of the inspected object, and each of the gray scales of the first detection probability and the first appearance probability are calculated. A first sum, which is the sum of the products for each tone, is calculated over all the above-mentioned gradations, and a second sum, which is the sum of the first appearance probabilities of each of the above-mentioned gradations, is calculated over all the above-mentioned gradations. the first total is divided by the second total, and the appearance probability of image data having the gray scale of the object to be inspected is calculated based on the image data of the object to be inspected. and the visual inspection device detects images of each degree of the feature amount based on image data including all degrees of the feature amount for each degree of the feature amount of the image representing the feature of the image of the inspected object. calculate a second detection probability that is the probability that the defect will appear; calculate a second appearance probability that is the probability that the defect will appear for each degree of the feature amount based on the image data of the inspected object; The distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has is calculated based on the detection probability of 2 and the second appearance probability, and the density level of the object to be inspected is calculated. The above-mentioned division value is added to the product of the appearance probability of image data having a tone and the above-mentioned distribution probability, and the above-mentioned added value is used as an evaluation value of the inspection ability of the above-mentioned appearance inspection device. Testing ability evaluation method. (2) In an inspection ability evaluation device for a visual inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, all a first calculation means for calculating a first detection probability that is a probability that the appearance inspection device can detect an image of each density gradation based on image data including the density gradation; a second calculation means for calculating a first appearance probability that is a probability that the defect will appear based on image data of the object to be inspected having a surface defect for each key; and the first detection probability and the above. a third calculation means for calculating a first sum, which is the sum of the products of each of the gray scales with the first appearance probability, over all of the gray scales; a fourth calculation means for calculating a second sum, which is the sum of the first appearance probabilities, over all the gray levels; and a fifth calculation means for dividing the first sum by the second sum. and a sixth step of calculating the appearance probability of image data having gray scale of the object to be inspected based on the image data of the object to be inspected.
and the visual inspection device calculates each degree of the feature amount based on image data including all the degrees of the feature amount of the image representing the feature of the image of the object to be inspected. a seventh calculation means for calculating a second detection probability, which is the probability that the image can be detected; and a second calculation means, which is the probability that the defect will appear for each degree of the feature amount based on the image data of the object to be inspected. and a degree of the feature amount other than the degree of the feature amount that the image data of the inspected object has based on the second detection probability and the second appearance probability. a ninth calculation means for calculating the distribution probability of the object to be inspected; and a tenth arithmetic means for adding, and the added value obtained by the tenth arithmetic means is used as an evaluation value of the inspection ability of the appearance inspection apparatus. Device. (3) In an inspection capability evaluation method for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, all A first detection probability, which is a probability that the appearance inspection device can detect an image of each density gradation, is calculated based on the image data including the density gradation, and detects a defect on the surface for each density gradation. A first appearance probability, which is the probability that the defect will appear, is calculated based on data of an image of the object to be inspected, and calculates each of the first detection probability, first appearance probability, and each other for the object to be inspected. A first sum, which is the sum of the products of each of the gray scales with a predetermined first importance level for each gray scale, is calculated over all the gray scales, and A second sum, which is the sum of the products of the appearance probability and the first importance level for each of the gray scales, is calculated over all the gray scales, and the first sum is calculated as the sum of the products of the first importance level and the first importance level.
, and calculate the appearance probability of image data having the gray scale of the object to be inspected based on the image data of the object to be inspected, Calculate a second detection probability that is the probability that the appearance inspection device can detect an image with each degree of the feature amount based on image data including all the degrees of the feature amount for each degree of the object to be inspected. A second appearance probability, which is the probability that the defect appears for each degree of the feature amount, is calculated based on the image data of the object, and the second detection probability, the second appearance probability, and the object to be inspected are calculated. Calculate the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on a second degree of importance determined in advance for each degree of the feature amount. , the above-mentioned division value is added to the product of the appearance probability of the image data having gradations of gradations of the above-mentioned object to be inspected and the above-mentioned distribution probability, and the above-mentioned added value is used as an evaluation value of the inspection ability of the above-mentioned visual inspection apparatus. Inspection ability evaluation method for featured appearance inspection equipment. (4) In an inspection ability evaluation device for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, all a first calculation means for calculating a first detection probability that is a probability that the appearance inspection device can detect an image of each density gradation based on image data including the density gradation; a second calculation means for calculating a first appearance probability that is a probability that the defect will appear based on image data of the object to be inspected having a surface defect for each key; and the first detection probability and the above. The first sum, which is the sum of the products of the first probability of appearance and the predetermined first importance level for each gray scale for the object to be inspected, for each gray scale is calculated as follows: a third calculation means that calculates across the gray scale; a fourth calculation means for calculating over the gray scale; a fifth calculation means for dividing the first sum by the second sum; and a fifth calculation means for dividing the first sum by the second sum. Sixth step for calculating the probability of appearance of image data having grayscale gradation
and the visual inspection device calculates each degree of the feature amount based on image data including all the degrees of the feature amount of the image representing the feature of the image of the object to be inspected. a seventh calculation means for calculating a second detection probability, which is the probability that the image can be detected; and a second calculation means, which is the probability that the defect will appear for each degree of the feature amount based on the image data of the object to be inspected. an eighth calculating means for calculating the appearance probability of the second detection probability, the second appearance probability, and a second importance predetermined for each degree of the feature amount for the object to be inspected; ninth calculation means for calculating the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on the degree of density; tenth calculation means for adding the division value calculated by the fifth calculation means to the product of the probability of occurrence of data and the distribution probability; An inspection ability evaluation device for a visual inspection device, characterized in that the evaluation value is an evaluation value of the inspection ability of the visual inspection device. (5) In an inspection capability evaluation method for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, A first appearance probability, which is a probability that the defect will appear, is calculated based on data of an image of the object to be inspected, which is determined in advance, and the appearance inspection device can detect images of each density gradation. A first sum, which is the sum of the products of the first detection probability, which is a probability, and the first appearance probability, for each of the gray scales, is calculated over all the gray scales, and A second sum, which is the sum of the first appearance probabilities of gradations, is calculated over all the gradation levels, and the first sum is divided by the second sum, and the image of the object to be inspected is calculated. The probability of appearance of image data having gradations of gradations of the object to be inspected is calculated based on the data, and each of the feature quantities of the image representing the image characteristics of the object to be inspected is calculated based on the image data of the object to be inspected. A second appearance probability, which is the probability that the defect will appear for each degree, is calculated, and a second detection probability, which is a predetermined probability that the appearance inspection device can detect an image with each degree of the feature amount, and the second Calculate the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on the appearance probability of 2, and calculate the appearance of image data having the gray scale of the object to be inspected. An inspection ability evaluation method for an appearance inspection apparatus, characterized in that the above-mentioned division value is added to the product of the probability and the above-mentioned distribution probability, and the added value is used as an evaluation value of the inspection ability of the above-mentioned appearance inspection apparatus. (6) In an inspection ability evaluation device for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, the surface is inspected at each predetermined gray level. a first calculation means that calculates a first appearance probability that is a probability that the defect will appear based on data of an image of the object to be inspected that has a defect; Calculate a first summation, which is the sum of the products of the first detection probability, which is the probability of detecting the gray scale image, and the first appearance probability, for each of the gray scale levels, over all the gray scale levels. a second calculation means for calculating a second sum, which is the sum of the first appearance probabilities for each of the gray scales, over all the gray scales; a fourth calculation means for dividing the total sum by the second summation; and a fifth calculation means for calculating the appearance probability of image data having the gray scale of the object to be inspected based on the image data of the object to be inspected.
a calculation means for calculating a second appearance probability, which is the probability that the defect appears for each degree of the feature amount of the image representing the feature of the image of the object to be inspected, based on the image data of the object to be inspected; a sixth calculation means, and an image of the object to be inspected based on the second detection probability and the second appearance probability, which are predetermined probabilities that the appearance inspection device can detect images of each degree of the feature amount. seventh calculating means for calculating the distribution probability of the degree of the feature amount other than the degree of the feature amount that the data has; and eighth calculation means for adding the division value calculated by the fourth calculation means to the product,
An inspection ability evaluation device for a visual inspection device, characterized in that the added value obtained by the eighth calculation means is used as an evaluation value of the inspection ability of the visual inspection device. (7) In an inspection capability evaluation method for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, A first appearance probability, which is a probability that the defect will appear, is calculated based on data of an image of the object to be inspected, which is determined in advance, and the appearance inspection device can detect images of each density gradation. The sum of the products of the first detection probability, which is a probability, the first appearance probability, and the predetermined first importance level for each gray scale for the object to be inspected, for each gray scale. A first sum that is calculated over all the gray scales, and a second sum that is the sum of the products of the first appearance probability and the first importance level for each gray scale. is calculated over all the gray scales, the first sum is divided by the second sum, and image data having gray scales of the object to be inspected is obtained based on the image data of the object to be inspected. Calculate the appearance probability of the defect, and calculate the second appearance probability, which is the probability that the defect will appear, for each degree of the feature amount of the image representing the feature of the image of the object to be inspected, based on the image data of the object to be inspected. A second detection probability, which is a predetermined probability that the appearance inspection device can detect an image with each degree of the feature amount, and the second appearance probability, and each of the feature amounts for the object to be inspected are calculated. Calculate the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has on the basis of a second degree of importance predetermined for each degree, and calculate the density of the object to be inspected. A visual inspection device characterized in that the division value is added to the product of the probability of appearance of image data having gradations and the distribution probability, and the added value is used as an evaluation value of the inspection ability of the visual inspection device. Testing ability evaluation method. (8) In an inspection ability evaluation device for an appearance inspection device that photographs an object to be inspected and inspects defects on the surface of the object based on the data of the photographed image, the surface is inspected at each predetermined gray level. a first calculation means that calculates a first appearance probability that is a probability that the defect will appear based on data of an image of the object to be inspected that has a defect; a first detection probability that is a probability of detecting a gray scale image; a first appearance probability; and a predetermined first importance level for each gray scale for the object to be inspected; a second calculation means for calculating a first sum, which is a sum of products for each tone, over all the grayscale levels; and each grayscale level of the first appearance probability and the first importance level. a third calculation means for calculating a second sum, which is a sum of products for each key, over all the grayscale levels; and a fourth calculation means for dividing the first sum by the second sum. , a fifth step of calculating the appearance probability of image data having gray scale of the object to be inspected based on the image data of the object to be inspected;
a calculation means for calculating a second appearance probability, which is the probability that the defect appears for each degree of the feature amount of the image representing the feature of the image of the object to be inspected, based on the image data of the object to be inspected; a sixth calculation means, a second detection probability that is predetermined and is a probability that the appearance inspection device can detect an image of each degree of the feature amount, the second appearance probability, and the above for the object to be inspected; a seventh operation for calculating the distribution probability of the degree of the feature amount other than the degree of the feature amount that the image data of the object to be inspected has based on a second degree of importance predetermined for each degree of the feature amount; means, and an eighth calculation means for adding the division value calculated by the fourth calculation means to the product of the appearance probability of the image data having the density gradation of the object to be inspected and the distribution probability. An inspection ability evaluation device for a visual inspection device, characterized in that the added value calculated by the eighth calculation means is used as an evaluation value of the inspection ability of the visual inspection device. (9) Claims 1 and 3, wherein the feature amount of the image is the number of pixels corresponding to the area of a dirt defect on the surface of the object to be inspected or the circumference length of the object to be inspected; 5. An inspection ability evaluation method for the appearance inspection device according to item 5 or 7. (10) Claims 2 and 4, wherein the feature amount of the image is the number of pixels corresponding to the area of a dirt defect on the surface of the object to be inspected or the circumference length of the object to be inspected; 6th
8. An inspection ability evaluation device for the appearance inspection device according to item 1 or item 8.