JPH08193959A - Device and method for detecting surface scratch of inspection object - Google Patents

Abstract

PURPOSE: To surely judge a scratch including a small scratch of an inspection object in a surface scratch detection device using an image processing device using brightness histogram. CONSTITUTION: A surface scratch detection device is provided with a brightness measuring part 6 for measuring the brightness level for each picture element of image data, an image memory part 7 for recording the brightness level for each picture element as picture element data, a plurality of brightness setting parts 12, 13, and 14 for setting and storing the range of a brightness level in advance, a plurality of brightness counting parts 9, 10, and 11 for calculating the number of picture elements within the range of a brightness level stored at each of a plurality of brightness setting parts out of the picture element data inputted from the image memory part, and a judgment part 15 for judging the presence or absence of the surface scratch of an inspection target based on the number of picture elements which are calculated by each of a plurality of brightness counting parts.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and method for detecting surface scratches on an inspection object using an image processing apparatus which can easily determine the presence or absence of surface scratches on the inspection object.

[0002]

2. Description of the Related Art An image processing apparatus has been researched and developed since early as a functional device acting as a substitute for human vision, and has finally reached a practical range. This is largely due to the improvement of processing performance due to the progress of hardware as well as the development of image processing algorithms. Image processing devices are considered to have a very wide range of application fields from the viewpoint of acting as a substitute for human vision, but from the viewpoint of the reliability of processing results, the fields that are actually put into practical use are extremely limited at present. ing.

In the field which is considered to be the most difficult in terms of reliability of the processing result, there is one generally called visual inspection. There are a method for detecting chipping of a corner portion of an inspection object such as a sheet material and a square material, and a method for detecting surface scratches on the inspection object. In the former case, if a known binarization process is used to process a region where the inspection target has a defect as a region where the inspection target does not exist, the binarized data of the pixel does not have a defect. The presence / absence of a defect can be detected relatively easily because the difference greatly depends on the presence / absence. However, in the case of the latter, it is difficult to uniquely define the threshold value that is the criterion for determining the presence or absence of scratches because the surface scratches have various shapes and sizes, and binarization as in the former case is difficult. Detection methods using processing cannot be easily applied.

For this reason, the detection of surface scratches on an inspection object using an image processing apparatus is currently put to practical use in very limited applications such as the detection of large scratches, but the detection of small scratches is performed by an inspector. It depends largely on the visual inspection of. However, the actual criteria for the visual inspection by the inspector may vary depending on the inspector, and even the same inspector may vary depending on the degree of fatigue. For this reason, it is indispensable to automate the manufacturing process including the inspection process by providing an apparatus capable of automatically detecting a small flaw without relying on the visual inspection by an inspector.

One of the most influential methods currently known as a method for automatically detecting surface flaws on an inspection object is:
A method of calculating the standard deviation of the brightness / darkness (luminance) histogram from the pixel data indicating the brightness of each pixel of the image of the inspection object, that is, the brightness level, and determining the presence or absence of surface scratches by the size of this standard deviation is available. is there. For example, in the magazine “Mechanical Design” (Vol. 37, No. 12, September 1993, pp. 63 to 65), a method for detecting surface flaws on the inner circumference of a bearing outer ring and an inner ring using a luminance histogram is disclosed. Is listed.

According to this, the fact that the brightness level in the area where surface scratches exist is lower than the brightness level in the area where surface scratches do not exist, and a sufficient number of bearings having no surface scratches on the inner circumference in advance are used. The luminance level of each pixel of the photographed image is obtained, the luminance histogram of the non-defective product is calculated from the average value of the luminance levels, and the standard deviation σ and the average value μ of the luminance level are calculated from the luminance histogram of the non-defective product, which is the inspection target. The brightness level on the inner circumference of the bearing is the reference value (μ
If there is an area smaller than -3σ), it is determined that there is a scratch. This method has an advantage that it is easy to put it into practical use because it is easy to set the judgment standard and the device constituting it is relatively simple.
When this method is used, when the surface of the inspection object is close to a mirror surface, the luminance distribution of the pixels becomes almost uniform, so the standard deviation σ of the reference luminance histogram becomes small, and the judgment reference value ( Since the distribution of the luminance histogram appears in a region smaller than (μ-3σ), it is possible to easily detect the surface flaw.

[0007]

However, since the surface of an actual metal-worked product including a bearing is usually finished by grinding, the surface looks fine, but in reality, many grinding patterns are present. If the above-mentioned method is used for metal processed products, the standard deviation σ of the reference luminance histogram becomes large, and there is a problem that the distribution of the luminance histogram does not appear in an area smaller than the judgment reference value (μ-3σ) with a small scratch. It was In other words, small scratches do not have a large difference from the ground pattern on the distribution of the luminance histogram, so that it was difficult to set a judgment standard for deriving a good judgment result by the above method.

An object of the present invention is to solve the above-mentioned problems of the prior art and to set a criterion for determining the presence / absence of surface scratches including small scratches in the detection of surface scratches on a metal processed product whose surface is finished by grinding. It is an object of the present invention to provide a surface flaw detection apparatus for an inspection target using an image processing apparatus and a surface flaw detection method using the same, which can easily perform the above.

[0009]

Therefore, the present invention provides a surface flaw detection apparatus using an image processing apparatus for detecting the surface flaw of an inspection object from the distribution of the luminance level of an image input from a camera or the like. A camera input terminal for inputting an inspection object image captured by a camera that outputs analog image data, an A / D converter for converting the inspection object image into digital image data, and each pixel of the digital image data Input from the image memory unit, a brightness measuring unit that measures the brightness level of each pixel, an image memory unit that records the brightness level of each pixel as pixel data, a plurality of brightness setting units that preset and store the range of brightness levels. Each of the plurality of brightness counters for calculating the number of pixels within the range of the brightness levels stored in each of the plurality of brightness setting parts of the pixel data To provide a surface flaw detection device for an inspection object, which has a determination unit for determining the presence or absence of a flaw on the inspection object based on the calculated pixel number and outputting the determination result to an output terminal. Solves the problems of the prior art.

Further, when digital image data is input to the camera input terminal, the digital image data is directly input to the luminance measuring section without passing through the A / D converter.

In order to determine the presence or absence of surface scratches on the inspection object, each of the brightness setting sections has a first brightness level range indicating the range of the brightness level of the background area of the inspection object, in advance.
A second brightness level range indicating a range of brightness levels of a non-scratch portion of the inspection object area, and a third brightness level range indicating a brightness level range of a scratched portion of the inspection object area. Are set individually.

With respect to the number of pixels included in the above-mentioned first luminance level range, the determination unit calculates the ratio (B /) of the number of pixels (B) of the inspection object to the number of pixels (A) of the reference object.
A) is 1 or more and is the first preset as the criterion.
Ratio (E /) of the number of pixels (E) included in the third brightness level range to the number of pixels (D) included in the second brightness level range that is equal to or less than the threshold value (C) of D)
However, when it is less than or equal to the second threshold value (F) set in advance as the determination standard, it is determined that the inspection target has no scratch.

[0013]

The number of pixels of the reference object (A) and the number of pixels of the inspection object (B) included in the first luminance level range set as the background portion of the reference object and the inspection object, and , The number of pixels (D) included in the second luminance level range that is set as a non-scratched portion of the inspection object area, and the third number that is set as a scratched portion of the inspection object area. The number of pixels (E) included in the brightness level range is calculated. The ratio (B / A) of the number of pixels (B) of the background portion of the inspection object to the number of pixels (A) of the background portion of the reference object is 1 or more and is set as a first criterion in advance. The ratio (E) of the number of pixels (E) of the scratched portion of the region of the inspection target to the number of pixels (D) of the region of the inspection target that is less than or equal to the threshold (C) and is not damaged. / D) is less than or equal to a second threshold value (F) preset as a determination criterion, it is determined that the inspection target has no scratch.

In other words, the ratio (B / A) of the number of pixels (B) in the background portion of the inspection object to the number of pixels (A) in the background portion of the reference object is 1 or less, or in advance as a criterion for determination. When it is equal to or larger than the set first threshold value (C),
It is determined that the inspection object is different from the reference object, and further, with respect to the number of pixels (D) of the non-scratched portion of the inspection object area, When the ratio (E / D) of the number of pixels (E) is less than or equal to a second threshold value (F) preset as a determination standard, it is determined that the surface of the inspection target has a scratch of an allowable value or more. To do.

[0015]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic view of an inspection system including a surface flaw detection device using an image processing device to which the present invention is applied. The inspection object 4 is photographed by the camera 1, this image is taken into the image processing device 2 and processed, and the determination result is displayed on the monitor 3. FIG. 2 is a block diagram showing the configuration of the image processing apparatus 2. The analog image data output from the camera 1 is input from the camera input terminal 21 of the image processing device 2, this analog image data is converted into digital image data by the A / D converter 5, and this digital image data is input. Is converted into pixel data converted into a brightness level for each pixel by the brightness measuring unit 6. When the camera 1 outputs digital image data, the digital image data input from the camera input terminal 21 is directly input to the brightness measuring unit 6 without passing through the A / D converter 5. You can do it like this.

The brightness level of the image data input to the brightness measuring unit 6 is measured for each pixel, and the brightness level is stored in the image memory 7 as pixel data. Image memory 7
Receives the memory address signal generated by the address generator 8 and sequentially outputs the stored pixel data. The pixel data sequentially output from the image memory 7 is input to the first brightness counting unit 9, the second brightness counting unit 10, and the third brightness counting unit 11. A first brightness setting unit 12, a second brightness setting unit 13, and a third brightness setting unit 14 are provided in pairs with the brightness counting units 9, 10, and 11. Each brightness setting unit 1
The upper limit value and the lower limit value of the brightness level are preset in 2, 13, and 14 as the range data of the brightness level of the pixel, and the brightness counting units 9, 10 to which the range data of the brightness level respectively correspond. , 11 are output. After all, in each of the brightness counting sections 9, 10 and 11, the pixel data sequentially output from the image memory 7 and the brightness setting sections 12 provided in pairs with the brightness counting sections 9, 10 and 11, respectively. , 13 and 14 respectively, the brightness level range data output from each of them is input simultaneously.

In each of the brightness counters 9, 10 and 11, the pixel data sequentially input from the image memory 7 and the brightness level range data input from the brightness setting units 12, 13 and 14 are compared, If the brightness level, which is pixel data, is within the range between the upper limit value and the lower limit value of the brightness level range data, the count-up is performed, and if it is outside the range, the count-up is not performed. After the comparison processing is performed on all the pixel data stored in the image memory 7, the number of pixels calculated by each of the brightness counting units 9, 10 and 11 is output to the determining unit 15.

Finally, the judging section 15 judges the presence or absence of scratches based on the number of pixels calculated in each of the luminance counting sections 9, 10, 11 and outputs the judgment result as judgment data to the output terminal 22.
It is designed to output to. The determination data output from the output terminal 22 is input to the monitor 3 and displayed. The determination data is fed back to the processing machine that processed the inspection object 4 (not shown) to perform abnormal processing such as stopping the operation of the processing machine. You may make it control.

Next, referring to FIG. 3 to FIG.
The method for determining the presence or absence of surface scratches on the inspection object performed in 1 will be described in detail. FIG. 3 shows an image obtained by photographing the end surface portion of the bearing outer ring which is the inspection object. The shaded portion indicates the end face portion 16, and the luminance level is higher than that of the region other than the shaded portion, that is, the background portion 17. Here, a brightness histogram is calculated for the entire screen. The brightness histogram is a graph in which the horizontal axis represents the brightness level and the vertical axis represents the number of pixels, and the number of pixels belonging to an arbitrarily set minute brightness level range is calculated. Here, the brightness level is the brightness of the pixel (brightness) of 25
It is shown in 6 gradation levels, and the brighter the pixel, the higher the brightness level. The brightness level is not always 2
It is not necessary to set 56 gradations, and higher gradations such as 512 gradations or 1024 gradations may be used. However, if the gradations are increased, the image processing time becomes longer. It is only necessary to select the minimum necessary gradation after considering the type and measurement accuracy.

FIG. 4 shows a luminance histogram when there is no scratch other than the grinding pattern on the end face portion 16. In this case, the luminance histogram is a peak 31 of the background portion in a region where the pixel luminance level is low (dark). Appears, and the peak 3 of the end face part appears in the area where the pixel brightness level is high (bright).
2 will appear. The width of the peak 32 of the end face portion that appears in the region of high brightness level is narrow for the inspection object with few grinding patterns, but is slightly wider for the inspection object with many grinding patterns.

Further, among the pixels of the end face portion 16, the luminance level of the pixel in the scratched portion becomes lower than the luminance level of the pixel in the non-scratched portion, so that the luminance histogram of the luminance histogram when the scratches exist in the end face portion 16 As shown in FIG. 5, the distribution is such that a peak 35 of the scratched end face portion appears at a position where the brightness level is lower than the peak 34 of the scratch free end face portion. Generally, the larger the scratched area of the end face portion 16, the higher the number of pixels of the peak 35 in the scratched end face portion, and vice versa.
The number of pixels of the peak 34 in the end surface portion without scratch is lower than the number of pixels of the peak 32 in the end surface portion of the luminance histogram in the case of no scratch.

From the characteristics of the brightness histogram when there is a scratch on the surface of the inspection object described above, in this embodiment, three distinct brightness level ranges are set in the brightness histogram,
The sum of pixels having a brightness level corresponding to the brightness level range will be calculated. The three characteristic brightness level ranges are a brightness level range of the background portion 17, a brightness level range of a non-scratch portion of the end face portion 16, and a brightness level range of a scratched portion of the end face portion 16. The three characteristic brightness level ranges include three brightness setting sections 1.
It is set at 2, 13, and 14, respectively. And
The three brightness counters 9, 10 and 11 are provided with the three brightness setting units 12, 13 and 14 which are installed in pairs.
The number of pixels belonging to the brightness level range set in is detected. That is, in the first brightness counting unit 9, the number of pixels belonging to the brightness level range of the background portion 17, and in the second brightness counting unit 10 the number of pixels belonging to the brightness level range of the non-scratch portion of the end face portion 16 is further increased. The third brightness counting unit 11 detects the number of pixels belonging to the brightness level range of the scratched portion of the end face portion 16, respectively.

FIG. 6 shows the luminance histogram when the end face portion 16 has a flaw, and the pixels detected by the respective luminance counting units of the first luminance counting unit 9, the second luminance counting unit 10, and the third luminance counting unit 11. The numbers are indicated by the shaded areas. In the present embodiment, the respective brightness ranges indicated by the brightness levels of 256 gradations are the brightness levels of 20 to 20 in the first brightness setting unit 12 as the first brightness level range as the brightness level range of the background portion 17.
In the second brightness setting unit 13, the second brightness setting unit 13 sets the second brightness level range as the brightness level range of the scratch-free portion of the end face portion 16 to the brightness levels 150 to 200, and the third brightness setting unit 1
4, the third luminance level range as the luminance level range of the scratched portion of the end face portion 16 is set to the luminance level 100 to
It was set to 120 respectively. However, the grinding pattern is not included in the third brightness level range. In other words, in the setting of the third brightness level range, it is necessary to set the upper limit value and the lower limit value so as to include pixels that may be determined as scratches but not pixels that are not determined as scratches such as a grinding pattern. There is.

FIG. 8 shows a flow chart of a method for judging the presence or absence of surface flaws on an inspection object in the judging section in the embodiment of the present invention. The number of pixels calculated by each of the brightness counting units 9, 10, 11 is output to the determining unit 15. The determination unit 15 first compares the number of pixels belonging to the background portion 17 detected by the first luminance counting unit 9 with the number of pixels in the background portion of the reference object. Here, the reference object refers to an object whose surface has almost no ground pattern, such as when the end surface portion 16 is mirror-finished. That is, the ratio (B / A) of the number of pixels (B) belonging to the background portion 17 of the inspection object 4 to the number of pixels (A) of the background portion of the reference object is based on a preset threshold value (C). If it is larger or smaller than 1 (step 42N), it is determined that the inspection object 4 has an abnormality and an abnormality signal is output to the output terminal 22 (step 4).
5).

This is because the number of pixels (B) of the background portion 17 of the inspection object 4 is not affected by the presence or absence of scratches on the end face portion 16 and the size of the scratches. If the number of pixels is abnormally larger than the number (A) of
This is because it is possible that the inspection object 4 has a defect other than a flaw such as a chip, a different inspection object, or no inspection object.
Further, when the ratio (B / A) is smaller than 1, that is, when the number of pixels (B) of the background portion 17 of the inspection object 4 is smaller than the number of pixels (A) of the background portion of the reference object, it is different. Since it is possible that the object is an object to be inspected, it is determined that the object is abnormal also in this case.

Next, when it is determined that the number of pixels belonging to the background portion 17 detected by the first luminance counting section 9 is normal (step 42Y), the flaw detected by the second luminance counting section 10 is detected. The ratio (E /) of the number of pixels (D) of the peak 34 of the non-end face portion to the number of pixels (E) of the peak 35 of the end face portion having a flaw detected by the third luminance counting unit 11.
D) is calculated, and if this ratio (E / D) is larger than a preset threshold value (F) (step 43N), an abnormal signal is output to the output terminal 22 as if the surface of the inspection object has a flaw. (Step 45). Further, the larger the scratches, the smaller the number of pixels (D) of the peak 34 at the non-scratched end face portion and the larger the number of pixels (E) of the peak 35 of the scratched end face portion. Therefore, the ratio (E / D ) Changes greatly, and even small scratches can be easily detected.

The determination method in the determination unit 15 of the present embodiment is intended to more specifically show the function of the determination unit of the present invention, and is not limited to this method. In this embodiment, three sets of the brightness counting unit and the brightness setting unit are set,
It is also possible to set four or more pairs within the scope of the present invention.
For example, in addition to the three sets of the brightness counting unit and the brightness setting unit that detect the three brightness ranges set in this embodiment, one set of the brightness counting unit and the brightness setting unit that detects the brightness range 80 to 100 may be provided. You may

Further, instead of individually setting the upper limit value and the lower limit value of the three sets of brightness level ranges as in the present embodiment, as shown in FIG. You may make it set the boundary value of an area | region and set three brightness level ranges. For example, if the boundary values of two adjacent luminance areas are set to 50 and 150, the three luminance level ranges are 0 to 50 for the background portion and 150 to 25 for the end surface portion without scratch.
6. The scratched end face portion is set to 50 to 150.

[0029]

According to the present invention, a surface flaw detection apparatus and method using an image processing apparatus for detecting surface flaws on an object to be inspected from the distribution of the brightness level of an image input from a camera or the like are used. A ratio (E / D) of the number of pixels (E) in the scratched portion of the region of the inspection object to the number of pixels (D) in the non-scratched portion of the region of the object is preset as a determination criterion. And the second threshold value (F) or less,
Since it is determined that the surface of the inspection object has more than the allowable value, it is possible to detect the presence of surface scratches including small scratches when detecting the surface scratches of metal processed products whose surface is finished by grinding. It became easier to set the standard.

[Brief description of drawings]

FIG. 1 is a schematic diagram of an inspection system using an image processing apparatus to which the present invention is applied.

FIG. 2 is a block diagram showing a configuration of an image processing apparatus of the present invention.

FIG. 3 is a diagram showing an image obtained by photographing an end surface portion of an outer ring of a bearing which is an inspection object.

FIG. 4 is an example of a luminance histogram when there is no scratch on the end face portion.

FIG. 5 is an example of a luminance histogram when there is a scratch on the end face portion.

FIG. 6 is a diagram illustrating a pixel detected by each of the brightness counting units in the case where the upper limit value and the lower limit value of the brightness range are individually set and three groups of brightness ranges are set in the brightness histogram when the end surface part has a scratch. It shows the number.

FIG. 7 is a luminance histogram in the case where there is a scratch on the end face portion, which is detected by each luminance counting unit when the boundary values of two adjacent luminance areas in the luminance area are set and three sets of luminance ranges are set. The number of pixels is shown.

FIG. 8 is a flow chart of a method for determining the presence / absence of scratches on an inspection target in a determination unit according to the embodiment of the present invention.

[Explanation of symbols]

 1 Camera 2 Image Processing Device 4 Inspection Object 5 A / D Converter 6 Luminance Measuring Unit 7 Image Memory Unit 9, 10, 11 Luminance Counting Unit 12, 13, 14 Luminance Setting Unit 15 Judgment Unit 21 Camera Input Terminal 22 Output Terminal A number of pixels in the background portion of the reference object B number of pixels belonging to the background portion 17 of the inspection object 4 threshold value of the number of pixels in the background portion D number of pixels of the peak 34 of the end surface portion without scratch E end portion with scratch Number of pixels in peak 35 of F

Claims (4)

[Claims] 1. A surface flaw detection device for an inspection object using an image processing device for detecting the surface flaw of the inspection object from the distribution of brightness levels of an image input from a camera or the like, and outputs analog image data. A camera input terminal for inputting an inspection object image captured by a camera, an A / D converter for converting the inspection object image into digital image data, and a brightness level for each pixel of the digital image data A brightness measuring unit for measuring, an image memory unit for recording the brightness level of each pixel as pixel data, a plurality of brightness setting units for presetting and storing a range of brightness levels, and a pixel input from the image memory unit. A plurality of brightness counters for calculating the number of pixels within the range of brightness levels stored in each of the plurality of brightness setting parts of the data, and each of the plurality of brightness counters. An apparatus for detecting surface scratches on an inspection object, comprising: a judgment unit for judging the presence or absence of surface scratches on the inspection object based on the calculated number of pixels and outputting the judgment result to an output terminal. 2. When digital image data is input to the camera input terminal, the digital image data is directly input to the luminance measuring unit without passing through the A / D converter. The surface flaw detection device for an inspection object according to claim 1. 3. The plurality of brightness counting units, a first brightness counting unit that calculates the number of pixels included in a first brightness level range set as a background portion of the inspection target, and the inspection target. Second set as a non-scratch part of the area
A second brightness counting unit for calculating the number of pixels included in the brightness level range of the first and a second brightness counting unit for calculating the number of pixels included in the third brightness level range set as a scratched portion in the region of the inspection object. 3. The surface flaw detection device for an inspection object according to claim 1 or 2, wherein the surface luminance detection part is a 3 luminance counting part. 4. The ratio of the number of pixels (B) calculated by the first brightness counting unit to the number of pixels (A) included in the first brightness level range of a reference object in the determination unit ( B / A) is greater than or equal to 1 and is less than or equal to a first threshold value (C) preset as a determination criterion, and the third brightness counting unit with respect to the number of pixels (D) calculated by the second brightness counting unit When the ratio (E / D) of the number of pixels (E) calculated in step 2 is less than or equal to a second threshold value (F) preset as a determination reference, it is determined that the inspection object has no scratch. The surface flaw detection method for an inspection object using the surface flaw detection device according to claim 3.