JP2002140695A - Inspection method and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to simply and accurately detect detects including a fine defect also on an uneven density part. SOLUTION: An inspection device 1 gets a reference image obtained by picking up an image of a model having no defect prior to inspection, executes extension processing and reduction processing and stores obtained extended and reduced images in an image memory 3 built in the device 1. At the time of inspection a pixel brighter than a corresponding pixel on the extended image and a pixel darker than a corresponding pixel on the reduced image about the image to be inspected are extracted and the existence of a defect is discriminated from a defect extraction image obtained by integrating the extracted results.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the field of processing gray-scale image data by a computer, and particularly to an object such as a molded product or a printed character, where foreign matter is attached, chipped, burred, scratched. The present invention relates to a technology for inspecting whether a defect such as a defect has occurred using a grayscale image (hereinafter, simply referred to as an “image”) obtained by imaging an object.

[0002]

2. Description of the Related Art In this type of conventional inspection, prior to the inspection, an image obtained by imaging a model having no defect is registered as a reference image, and the difference between the image of the inspection object and the reference image is calculated. To extract the difference between the two images (inter-image calculation processing), or to calculate the similarity between the reference image and the inspection target image by the normalized correlation calculation.

[0003]

In general, in the case of a grayscale image, the phenomenon that the gradient of the density gradient near the outline fluctuates slightly for each image due to the positional relationship of the outline of the object with respect to the image sensor of the camera. Occurs. Therefore, no matter how exactly the image of the target object and the reference image are aligned, a difference occurs in the density near the edge, and when the inter-image calculation processing is performed, the difference is extracted as a defect. This causes a problem. Therefore, conventionally, when a difference between an inspection target image and a reference image is extracted by the inter-image calculation processing, an image near an edge is excluded from a processing target.

However, when such a process is performed, there is a problem that when a defect exists near an edge, the defect cannot be detected. Further, prior to the inspection, it is necessary to appropriately set an area to be excluded from the processing target. However, such setting is not easy and requires a complicated operation, which imposes a heavy burden on the user.

In the case of an object whose surface has irregularities,
The unevenness of the density of the object on the grayscale image occurs due to the unevenness. Therefore, when the inter-image arithmetic processing is applied to such an object, a difference in the concavo-convex pattern between the inspection object and the model is extracted as a defect, and a problem occurs that a defective portion cannot be accurately extracted. I do.

On the other hand, when the normalized correlation calculation is used, even if there is a minute defect in the inspection object, the difference due to the defect is not reflected in the correlation value, so that the minute defect cannot be detected with high accuracy. is there.

The present invention has been made in view of the above problems, and can easily and accurately detect a defect near an edge or a portion having uneven shading or a minute defect without troublesome setting operation. It is intended to realize such an inspection.

[0008]

In order to achieve the above object, according to the present invention, there is provided a case where a gray-scale image obtained by capturing an image of a predetermined object is taken into a computer and the presence or absence of a defect in the object is inspected. Prior to the inspection, the reference image obtained by imaging the model of the object having no defect is expanded and contracted, and the obtained dilated image and contracted image are registered in the memory. At the time of inspection, a difference between the dilated image and the contracted image is extracted from the grayscale image of the inspection object, and determination data for determining the presence or absence of a defect in the inspection object is generated from each extraction result. I have.

[0009] The term "shade image" as used herein means digital image data of a predetermined bit length converted from an analog image signal obtained by imaging an object by an analog camera.
Or image data generated by a digital camera. In addition, although a monochrome grayscale image is mainly used for this type of inspection, the present invention is not limited to this, and a color grayscale image may be used.

In expansion and contraction processing for a grayscale image, attention is paid to the constituent pixels of the image to be processed in order, and the brightness of the pixel of interest is determined according to the brightness of the pixel and the mask area including a plurality of pixels in the vicinity. To set. For example, in the dilation process, the brightness of the brightest pixel in the mask
Assuming that the brightness of the darkest pixel in the mask area is set to the pixel of interest in the erosion process, the image data corresponding to the shading unevenness on the reference image is stable in the bright direction in the expanded image, Becomes stable in the dark direction. Further, the brightness of the pixel near the edge on the image shifts in the bright direction in the expanded image and shifts in the dark direction in the contracted image.

When the dilation processing and the erosion processing are performed by the above-described method, in the processing for extracting the difference between the dilated image and the eroded image from the gray image of the inspection target, specifically, the inspection target image and the dilated image are extracted. This is performed by comparing the brightness of each of the contracted images with the corresponding pixels for each corresponding pixel, and extracting the pixels having a difference. In a preferred embodiment, a pixel brighter than the brightness of the corresponding pixel on the dilated image and a pixel darker than the brightness of the corresponding pixel on the eroded image in the inspection target image are extracted.

According to the above-described dilation processing, any pixel on the dilated image may be brighter than the original reference image but not darker. Therefore, the object on the inspection target image matches the model on the reference image,
If there is no defect, pixels brighter than the dilated image are not extracted in a portion where the density on the inspection target image is stable. Similarly, according to the contraction processing, any pixel on the contracted image may be darker than the brightness of the original reference image, but not brighter. Therefore,
If the object on the inspection target image matches the model on the reference image and has no defect, no pixel darker than the contracted image is extracted in a portion where the density on the inspection target image is stable. .

As described above, the density near the edge and the gradient of the density gradient are different for each image.
It is considered that the upper limit value and the lower limit value of the brightness near the edge are almost the same unless the illumination condition is changed. Therefore, if there is no defect near the edge on the inspection target image having a brightness outside the brightness range originally indicated, the brightness of each pixel near the edge is within the brightness range indicated by the dilated image and the contracted image. Are not included in the comparison with any of the images. On the other hand, when there is a defect having a brightness outside the original brightness range near the edge of the inspection target image, the brightness of the pixel of this defective portion is brighter than the brightness of the corresponding pixel on the dilated image,
Alternatively, the brightness is lower than the brightness of the corresponding pixel on the contracted image. Therefore, the defect can be extracted by comparing the image with any one of the expanded and contracted images.

[0014] Similarly, in the case where shading occurs due to unevenness on the surface or the like, the range in which the brightness varies in the shading portion is considered to be substantially the same unless the illumination conditions are changed. Therefore, if there is no defect having a brightness exceeding the range of the original brightness variation in the shading unevenness portion on the inspection target image, a difference is extracted in the shading unevenness portion in comparison with either the dilated image or the contracted image. Not done. on the other hand,
If there is a defect having a brightness outside the range of normal brightness variation in the shading unevenness portion of the inspection target image, this defect is extracted by comparison with an image of either expansion or contraction. Therefore, it is possible to accurately extract only a defect exceeding the range of the brightness variation without picking up noise due to the normal brightness variation on the image.

Further, according to the present invention, as an image processing apparatus for carrying out the above method, an image input means for capturing a grayscale image obtained by imaging a predetermined object, When a reference image obtained by imaging the model is captured, expansion / contraction processing means for expanding and contracting the reference image, a memory for storing an expansion image and a contraction image obtained by the expansion / contraction processing, Each time a gray image obtained by imaging the inspection object is input by the image input unit, a difference extraction unit that extracts a difference between the input image and the dilated image and the contracted image in the memory, and an extraction result by the difference extraction unit And a data generating means for generating determination data for determining the presence or absence of a defect in the target object.

The image input means is constituted by an interface circuit or the like for receiving an image signal from the image pickup means. In particular, when the image pickup means is constituted by an analog camera, an A / D conversion circuit and a noise cut filter are provided. Such a configuration is also included. The memory for storing the dilated image and the contracted image is configured by a RAM or a non-volatile memory. The expansion / contraction processing means, the difference extraction means, and the data generation means are realized by providing a computer with a processing program for each means. However, the present invention is not limited thereto, and an arithmetic processing circuit may be configured for each means. It is.

According to the apparatus having the above configuration, when a model of an object is imaged, an expanded image and a contracted image are automatically generated from the reference image generated by the imaging and registered in the memory. Next, in the inspection, every time a gray image of the inspection object is captured, a difference between the dilated image and the contracted image is automatically extracted from the image, and determination data for determining the presence or absence of a defect in the inspection object is created. You. Therefore, the user does not need to perform complicated processing for measurement, and it is possible to accurately detect a minute defect or a defect in a portion having uneven density.

According to a preferred aspect, the determination data is created as a defect extraction image in which the difference extraction results for the dilated image and the contracted image are integrated, and the defect extracted image is displayed on a monitor screen. For example, with respect to the inspection target image, when a pixel brighter than the brightness of the corresponding pixel on the dilated image and a pixel darker than the brightness of the corresponding pixel on the eroded image are extracted, the pixels extracted by these processes are extracted. By generating an image represented by a density value of a pixel or a density difference from a contrasted image, it is possible to create a defect extraction image representing a portion whose density is significantly different from the reference image. Further, it is also possible to create a defect extraction image that has been subjected to processing such as binarizing the density value of each extracted pixel or converting the density value to a constant density level. According to such processing, a defect of the inspection object can be accurately extracted and an image showing only the extracted defect can be displayed, so that even a minute defect can be clearly shown.

In still another preferred embodiment, it is determined whether or not the inspection object has a defect using the determination data, and a result of the determination is output. In the determination of the presence or absence of a defect, it is confirmed whether or not a difference has been extracted from the determination data with respect to the dilated image or the contracted image. . However, if the presence or absence of a defect is determined according to the size or density level of the pixel extracted as a difference, the level of detection accuracy can be set arbitrarily, such as removing a minute defect as noise. Inspection to determine the size and shape of a defect can be performed. With this configuration, each time an image to be inspected is input, the presence or absence of a defect in the object is automatically determined, and the result of the determination is output, so that the defect inspection can be completely automated.

[0020]

FIG. 1 shows the configuration of an inspection apparatus according to one embodiment of the present invention. This inspection device 1 is for inspecting the state of printed characters on a package and the like.
Image input unit 2, Image memory 3, Image output unit 4, Timing control unit 5, Character memory 6, Control unit 7, Monitor 8, I /
It is composed of an O port 9 and the like.

The image input unit 2 comprises an interface circuit for taking in a grayscale image signal from an analog camera (not shown), an A / D converter for digitally converting the grayscale image signal, a filter circuit for noise cut, and the like. You. The camera is not limited to an analog camera, but may be a digital camera.

The image memory 3 fetches and stores digital grayscale image data fetched by the image input unit 2, and stores a reference image, an expanded image, a contracted image, a defect extracted image, and the like, which will be described later. Have an area. The character memory 6 stores text data for displaying character information such as an inspection result and a display position thereof. Each of the memories 3 and 6 is connected to a control unit 7 via an address / data bus 10, and outputs data according to an instruction from the control unit 7 to an image output unit 4 according to a timing signal from the timing control unit 5. Alternatively, the data is output to the address / data bus 10.

The control unit 7 includes a CPU 11, a ROM 1
2, a hard disk 14 having a RAM 13 as a main body, and a control program indicating a processing procedure for inspection installed therein. Based on the control program in the hard disk 14, the CPU 11 reads and writes information from and to each memory via the address / data bus 10 and creates a defect extraction image indicating the presence or absence of a defect in the inspection object. Further, the CPU 11 determines the quality of the inspection object using the defect extraction image.

The monitor 8 is used to display the defect extraction image and the result of pass / fail judgment when a series of inspections is completed. The I / O port 9 is connected to an input unit such as a keyboard and a mouse, and to an output unit such as an external storage device and a transmission unit. The I / O port 9 inputs various setting data from the input unit and outputs inspection results to the outside. Used when

In this inspection apparatus, prior to the inspection, a non-defective non-defective model is imaged to generate a reference image, and the reference image is subjected to expansion processing and contraction processing.
The obtained dilated image and contracted image are stored in the image memory 3 together with the reference image. Then, at the time of inspection, every time an image of the inspection object is input, the difference between the inspection object image and the dilated image and the contracted image is extracted, and the defect extraction image is generated.

In the expansion and contraction processing, while a 3 × 3 mask as shown in FIG. 2 is scanned on the reference image, a central pixel g (coordinate (x, y in FIG. ), And the density value is converted to a predetermined value. In this embodiment, the density value is set so as to increase in the bright direction. In the expansion processing, the bright portion on the reference image is expanded by assigning the maximum density value in the mask to the target pixel g. Conversely, in the shrinking process,
By assigning the minimum density value in the mask to the target pixel g, the dark portion on the reference image is expanded.

FIG. 3 shows a specific example of a reference image and an expanded image and a contracted image generated from the reference image. In the case of the illustrated example, the character portion to be inspected is darker than the background portion. Therefore, in the dilated image, the background portion is enlarged to the character side, and a character thinner than the character in the reference image appears.
Conversely, in the contracted image, as a result of the character portion being enlarged toward the background, a character that is thicker than the character in the reference image appears.

FIG. 4 schematically shows the brightness of an image in one direction on the image. FIG. 4A shows an example in which shading does not occur inside and outside the edge. A in the figure is a change in brightness in the reference image, and an inclined portion that transitions from a bright portion to a dark portion corresponds to an edge. D in the figure
E indicates a change in brightness in the dilated image, and E indicates a change in brightness in the contracted image. As shown in FIG. 3, in the dilated image, the position of the edge shifts in a direction darker than the reference image, so that the brightness of the pixel representing the edge on the reference image shifts in a direction brighter than the reference image. I do. On the other hand, since the edge on the contracted image shifts in a direction brighter than the edge of the reference image, the brightness of the pixel representing the edge on the reference image shifts in a direction darker than the reference image.

FIG. 4B shows a change in brightness when shading occurs inside and outside the edge. In the figure, A1 represents a change in brightness in the reference image, D1 represents a brightness level enhanced by the dilated image, and E1 represents a brightness level enhanced by the eroded image.

As shown in FIG. 4 (2), when the brightness varies inside and outside the edge on the reference image, the result of replacing the density value of each pixel with the maximum density value in the mask in the expanded image. Then, the brightness of the portion of the shading unevenness becomes stable around the maximum level of the brightness variation. Conversely, for the contracted image, the density value of each pixel is replaced with the minimum density value in the mask, and as a result, the brightness of the shaded portion becomes stable near the minimum level of brightness variation.

Next, in the process of extracting the difference between the dilated image and the contracted image of the input image to be inspected in the inspection, specifically, the following equation (1) is used for the corresponding pixels between the images. Is performed, the density value of each of the constituent pixels H (x, y) of the defect extraction image is obtained. H (x, y) = G (F (x, y) -D (x, y)) + G (E (x, y) -F (x, y)) (1)

In the above equation (1), F (x, y) is
The density value of the pixel of interest of the inspection target image, D (x, y), is the density value of the pixel on the dilated image corresponding to the pixel of interest, E
(X, y) is the density value of the pixel on the contracted image corresponding to the pixel of interest. Note that the function G (X) is G when X> 0.
When (X) = X and X ≦ 0, G (X) = 0,
Is defined.

Here, focusing on the relationship between the inspection object image and the dilated image, in the dilated image, since the highest density value is set in the vicinity of each of the pixels, generally F (x, y) − The relationship of D (x, y) ≦ 0 holds. This relationship is the same for a portion in which shading occurs on the inspection target image, and the difference between the images is set to 0 by the function G. However, in the inspection target image, FIG.
If a pixel that is brighter than the original brightness range exists, as in B3 and B5 in (2), F
(X, y) -D (x, y)> 0, and the difference value is extracted by the function G.

Next, focusing on the relationship between the image to be inspected and the contracted image, in the contracted image, since the lowest density value is set in the vicinity of each of the pixels, generally E (x, y) − The relationship of F (x, y) ≦ 0 holds, and the difference between the images is set to 0 by the function G. However, if pixels that are darker than the original brightness range exist in the inspection target image, such as B4 and B6 in FIG.
For this pixel, E (x, y) -F (x, y)> 0
The difference value is extracted by the function G.

Further, the gradient of the density gradient at the edge portion of the image has a slight difference for each image. However, if the illumination conditions are the same, the variation range of the brightness of the edge and the pixels in the vicinity thereof is expanded. It can be considered to be included in the range of the brightness indicated by the image and the contracted image. Therefore, when the brightness of each pixel of the edge portion in the inspection target image is included in the range of the brightness of the edge portion on the reference image, F (x, y) −D (x, y) ≦ 0 and E
Since the relationship of (x, y) -D (x, y) ≦ 0 holds,
The difference between the images is set to 0 by the function G.

On the other hand, in the inspection target image, FIG.
If there is a pixel having a brightness outside the original variation range of the brightness of the edge portion like B1 and B2, F (x, y) −D (x, y)> 0 or E (x,
y) −F (x, y)> 0, and the difference value is extracted by the function G.

That is, according to the equation (1), among the constituent pixels of the image to be inspected, the pixels that are brighter than the corresponding pixels on the dilated image and the pixels that are darker than the corresponding pixels on the eroded image are the same as the corresponding pixels. The density difference is extracted, and the density value is replaced with 0 for the other pixels, so that a defect extraction image representing only the defect portion of the inspection object is generated. Also, in a portion where shading unevenness occurs, such as near an edge or an uneven portion, brightness included in the normal brightness fluctuation range is not extracted as a difference, but pixels brighter or darker than the normal brightness fluctuation range are not extracted. , It is possible to accurately extract a defective portion without adding noise due to uneven shading.

In addition to the processing according to the above equation (1), the average value of the gray level in the image to be inspected and the average value of the gray level in the reference image are obtained, and the value calculated in (1) is normalized by the average value of the gray levels. For example, it is possible to remove the influence of the difference in brightness between the reference image and the inspection target image, and perform a more accurate defect extraction process.

FIG. 5 shows a series of procedures in the inspection apparatus. First, in ST1, the model of the inspection object is imaged by the camera, and the input image is stored in the image memory 3 as a reference image. Next, in ST2, the dilation processing and the erosion processing described above are performed on the reference image. The dilated image and the eroded image obtained by each processing are stored in a predetermined area in the image memory 3 as in the case of the reference image.

After the completion of such preprocessing, the inspection object is sequentially sent to the image pickup position of the camera to perform the inspection. In the inspection, first, in ST3, an image of the inspection object is fetched and stored in the image memory 3. Then S
At T4, the position of the inspection target image is shifted with respect to the reference image by a measurement process after binarizing or extracting an edge of the inspection target image or a pattern matching process with the target object model extracted from the reference image. And the inclination are measured, and the positional deviation and the inclination are corrected.

In step ST5, the equation (1) is executed using the inspection object image, the dilated image, and the eroded image to generate a defect extraction image. In the following ST6,
The generated defect extraction image is provided to the image output unit 4 and displayed on a monitor screen.

FIG. 6A shows an example of an image to be inspected. In addition to an image of a character to be inspected, an image showing a minute defect appears. FIG. 6B shows a defect extraction image generated by the process of ST5 for the input target image. As described above, according to the expression (1), the pixel corresponding to the character of the background or the object on the input image is “0”, and only the pixel representing the defect is compared with the corresponding pixel on the dilated image or the eroded image. Is extracted. In FIG. 6 (2), the extraction result is shown in black and white inverted for easy understanding of the defective portion. However, the present invention is not limited to this, and the defect extraction image generated by Expression (1) is displayed as it is. It is also possible to binarize and show the defect extraction image.

Returning to FIG. 5, in ST7, it is determined whether or not there is a pixel indicating a defect on the defect extraction image.
The result of the determination is output on a monitor screen. After that, if the inspection is not completed, the process returns from ST8 to ST3, fetches the next image of the inspection object, and similarly executes the processing of ST4 to ST7.

In the determination process of ST7, the presence / absence of a defect may be simply determined. However, the difference value of the density or the size of the defect in the defective portion is compared with a predetermined threshold value to determine the defect. The degree may be determined with high accuracy. In addition, when a slight difference in density is extracted by the equation (1), it is desirable to determine a defect only when a difference value exceeding a threshold value is obtained so that the difference is not determined as a defect.

[0045]

According to the present invention, judgment data for judging the presence or absence of a defect in an inspection target image is generated by using an expansion image and a contraction image automatically generated from a reference image. Even a minute defect can be detected easily and accurately without any trouble. In particular, even in a portion where noise is easily extracted, such as near an edge or an uneven portion, it is possible to extract a defect having a brightness outside a normal fluctuation range of shading, and it is possible to detect the defect with high accuracy.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an inspection apparatus according to one embodiment of the present invention.

FIG. 2 is an explanatory diagram showing a mask used for expansion processing and contraction processing.

FIG. 3 is an explanatory diagram showing a reference image and an expanded image and a contracted image generated from the image.

FIG. 4 is an explanatory diagram showing a density distribution near an edge of an image.

FIG. 5 is a flowchart showing a processing procedure in the inspection device.

FIG. 6 is an explanatory diagram showing an inspection target image and a defect extraction image obtained by processing the image.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Inspection apparatus 2 Image input part 3 Image memory 7 Control part 11 CPU

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 2G051 AA90 AB02 AC04 CA04 ED15 ED30 5B057 AA01 BA02 CA08 CA12 CA16 CB08 CB12 CB16 CE09 DA02 DA07 DC22 DC32 5L096 BA03 EA02 FA17 GA08 GA41

Claims (8)

[Claims] 1. A method for inspecting the presence or absence of a defect in an object by taking a gray-scale image obtained by imaging a predetermined object into a computer, the method comprising: The reference image obtained by imaging is expanded and contracted, and the obtained dilated image and contracted image are registered in a memory, and for the gray image of the inspection object, for each of the dilated image and the contracted image, An inspection method characterized by extracting differences and generating determination data for determining the presence or absence of a defect in an inspection object from each extraction result. 2. A method of extracting a difference between a grayscale image and a dilated image of a test object with respect to an expanded image and a contracted image, the pixel being brighter than the corresponding pixel on the dilated image and the pixel being darker than the corresponding pixel on the contracted image. The inspection method according to claim 1, wherein the inspection method is a process of extracting the information. 3. The defect extraction image according to claim 1, wherein the determination data is a defect extraction image obtained by integrating a difference extraction result with respect to each of the dilated image and the contracted image. Inspection method described. 4. The inspection method according to claim 1, wherein the determination data is used to determine whether the inspection object has a defect, and a result of the determination is output. Inspection method. 5. An image input unit for capturing a grayscale image obtained by imaging a predetermined object, and a reference image obtained by imaging a model of a defect-free object by the image input unit. Expansion and contraction processing means for expanding and contracting the reference image, a memory for storing the expansion image and the contraction image obtained by the expansion and contraction processing, and an image of the inspection object captured by the image input means. Extracting means for extracting a difference between the input image and the dilated image and the eroded image in the memory each time the obtained grayscale image is input; and determining whether or not there is a defect in the object from the extraction result by the difference extracting means. An inspection apparatus comprising: data generation means for generating determination data for determination. 6. The method according to claim 5, wherein the difference extracting unit extracts a pixel brighter than a corresponding pixel on the dilated image and a pixel darker than the corresponding pixel on the contracted image from the grayscale image of the inspection object. Inspection device described. 7. The apparatus according to claim 5, wherein the data generation unit generates a defect extraction image in which a difference extraction result for each of the dilated image and the contracted image is integrated as the determination data. And a display control means for displaying the generated defect extraction image on a monitor screen. 8. The apparatus according to claim 5, wherein the determination unit determines whether the inspection object has a defect using the determination data generated by the data generation unit. And an output unit for outputting the determination result.