JP5784406B2 - Inspection method for processed products - Google Patents

Description

  The present invention relates to a method for inspecting a processed product for detecting burrs generated by processing and foreign matters placed on the processed product, and related techniques.

  As one of inspection methods for processed products, there is a method of photographing processed products and detecting processing defects such as burrs by image analysis (see Patent Documents 1 and 2).

  Patent Document 1 describes a method for inspecting protrusions such as burrs generated on a processed end face. The surface processed end face is photographed by a plurality of cameras from different angles, and the projections are compared by comparing these images. The presence or absence is detected.

  Patent Document 2 describes an inspection method in which air is blown onto a glass color filter which is a position to be inspected, and before and after the image is compared, dust and welded foreign matter are identified. . According to this inspection method, among the foreign matter candidates detected in the image before air blowing, those that do not exist after air blowing are determined to be dust blown off by air and not foreign matter, and after air blowing Also, it is determined that the existing foreign matter is a deposited foreign matter.

Japanese Patent Laid-Open No. 9-218952 Japanese Patent Laid-Open No. 10-246705

  However, there are minute irregularities that are not defects on the surface of the metal workpiece, and there are also minute irregularities due to surface processing such as hairline processing. Further, in a processed product obtained by processing a screw hole or the like on the inspection surface, a stepped portion appears in the image as a light / dark difference, or discoloration due to processing appears as a color tone difference. For this reason, in an image analysis based on an image having minute unevenness or a contrast of light and dark as a background, it is difficult to detect a long and narrow defect such as a burr, and erroneous determination may occur.

  In view of the above-described technical background, an object of the present invention is to provide an inspection method capable of accurately identifying defects such as burrs even in a processed product having minute irregularities on the surface.

  That is, this invention has the structure as described in following [1]-[7].

  [1] Comparing an in-air-blow image captured while air is blown to a processed product and an airless image captured without blowing air onto the processed product, and detecting a defect based on a form change caused by air blowing A method for inspecting processed products.

  [2] The method for inspecting a processed product as recited in the aforementioned Item 1, wherein the airless image is one or both of an image before air blow taken before air is blown and an image after air blow taken after air is blown.

  [3] The method for inspecting a processed product according to item 1 or 2, wherein a change in the form of a defect is detected based on a gradation difference between the image during air blowing and an image without air.

  [4] Form of defects by detecting a portion where the data of adjacent pixels in the air-blowing image and air-free image change rapidly, extracting the boundary of the change, and comparing the boundary extracted from the two images 3. The method for inspecting a processed product according to 1 or 2 above, wherein a change is detected.

  [5] The method for inspecting a processed product according to the item 1 or 2, wherein a defect shape change is detected by pattern matching between the image during air blowing and the image without air.

  [6] The method for inspecting a processed product according to any one of items 1 to 5, wherein the processed product has a screw hole as a processed part.

[7] An air blowing device that blows air on a processed product;
A CCD camera that shoots an image during air blowing while blowing air to the processed product, and shoots an airless image without blowing air to the processed product;
An apparatus for inspecting a processed product, comprising: an image analysis device that compares the image during air blowing with an image without air and detects a defect based on a change in form caused by air blowing.

  According to the invention described in [1] above, by comparing the air-in-air image with the air-free image, air is blown against minute irregularities on the surface of the processed product that does not move even when air is blown and does not change in form. It is possible to identify defects such as burrs and foreign matters whose shapes change, and to inspect the presence or absence of these defects in a processed product.

  According to the invention described in [2] above, the above-described effect can be obtained by comparing one or both of the pre-air blow image and the post-air blow image taken after air blowing as an airless image with the air blowing image.

  According to the invention described in [3] above, it is possible to detect a change in the form of a defect based on a gradation difference between the image during air blowing and the image without air, and obtain the above-described effect.

  According to the invention described in [4] above, in the air blowing image and the airless image, a portion where the data of adjacent pixels changes rapidly is detected, the boundary of the change is extracted, and extracted from two images. By comparing the boundaries, a change in the shape of the defect can be detected, and the above effect can be obtained.

  According to the invention described in [5] above, it is possible to detect a change in the form of a defect by pattern matching between the image during air blowing and the image without air, and obtain the above-described effect.

  According to the invention described in [6] above, since it is possible to distinguish the background and the defect even in a processed portion in which a step due to a screw thread is formed in a stripe shape like a screw hole, Accurate defect detection is possible even in the inspection of burrs generated during processing.

  According to the workpiece inspection apparatus described in [7] above, the inspection method described in [1] to [6] can be performed to accurately inspect defects.

It is a perspective view which shows the outline of the inspection apparatus which enforces the inspection method of the workpiece of this invention. It is the image of the screw hole of a to-be-inspected object, and its vicinity. It is an image before air blow in the photographed image 1. It is an air blowing image in the photographed image 1. It is an image after air blow in the photographed image 1. It is an image before air blow in the photographed image 2. It is an air blowing image in the photographed image 2. It is the image of the screw hole of other to-be-inspected objects, and its vicinity. It is an image before air blow in the photographed image 3. It is an air blowing image in the photographed image 3. It is the image after air blow in the picked-up image 3. FIG. 4 is a diagram illustrating a gradation difference between FIG. 3A and FIG. 3B in inspection example 1; It is a figure which shows the gradation difference of FIG. 3B and FIG. 3C in the test example 2. FIG. 4A and 4B are diagrams illustrating a gradation difference between FIGS. 4A and 4B in Inspection Example 3. FIG. FIG. 6 is a diagram showing a gradation difference between FIG. 5A and FIG. 5B in inspection example 4; FIG. 5B is a diagram showing the gradation difference between FIG. 5B and FIG. 5C in Inspection Example 4. FIG. 5B is a diagram showing the gradation difference between FIG. 5A and FIG. 5C in Inspection Example 4.

  FIG. 1 is a perspective view showing an outline of an inspection apparatus for carrying out a workpiece inspection method according to the present invention.

  The inspection device (1) captures and stores the images taken by the CCD camera (2) and the CCD camera (2) for photographing the object (10) fixed on the examination table (not shown). It has an image analysis device (3) for analyzing the image, an oblique illumination (4) arranged on the same axis as the CCD camera (2), and an air blowing device (5) for blowing air to the object to be inspected (10) . In the air blower (5), only a nozzle (5a) that blows out air is illustrated, and the direction of blowing air can be freely changed by changing the position of the nozzle (5a). The illumination is not limited to the coaxial down-tilt illumination shown in the drawing, and any illumination such as ring illumination or general reflection illumination can be used.

  In the method for inspecting a processed product according to the present invention, a defect is detected by comparing an in-air-blow image captured while air is blown onto a processed product that is an object to be inspected and an airless image captured without blowing air. The airless image includes both the pre-air blow image taken before air blowing and the post-air blow image taken after air blowing, and at least one of the airless images is compared with the air blowing image. Defects to be detected include burrs that are protrusions connected to the object to be inspected (10), and chips that are not connected to the object to be inspected (10) and are placed on the object to be inspected. Or foreign matter such as dust.

  When air is blown to the burr, the burr being blown vibrates, or bends at the root or middle of the burr and changes direction or form, or bends and vibrates in a changed direction or form. There is movement, and the position and form change before air is blown. Since these changes return to the position and form before the air blowing when the air blowing is stopped, the burr during the air blowing is different from the state before or after the air blowing.

  In the photographed image, the burr appears as a light / dark difference or color difference (hereinafter abbreviated as “light / dark difference”) with respect to the background, and the form formed by the position where the light / dark difference appears or the light / dark difference, etc. There are no images and different. That is, the burr that is bent at the base or in the middle of the length direction is clearly different from the position and form of the burr in the airless image. Since the vibration area of the burrs being vibrated is photographed, the area and gradation in which the contrast between the background and the airless image obtained by photographing the non-vibration burrs are different.

  On the other hand, micro unevenness on the surface peculiar to metal processed products, micro unevenness due to surface processing such as hair alignment processing, and unevenness caused by tapping and drilling do not move even when air is blown, and the position and form do not change. These irregularities also appear as light and dark differences in the image like burrs, but the position and form of the light and dark differences do not change depending on whether air is blown or not. Therefore, by comparing the image during air blowing and the image without air, it is possible to identify only those having a change in position or form as burrs among those appearing as light and dark differences. In addition, the approximate size of the burr can be detected from the area where the change appears.

  In addition, foreign matters that are merely placed on the object to be inspected, such as chips and dust, are blown away when air is blown off, or are removed from the object to be inspected, or the position of the entire foreign material changes, so the position of the base does not change. It can be distinguished from Bali.

  Since the foreign matter is not present in the inspected object after the inspection, if it is not necessary to determine that the foreign matter was present before the inspection, it is removed by blowing air before the inspection. It is preferable. By removing foreign matters that do not need to be determined in advance, image analysis is simplified and more accurate inspection is possible. Moreover, since the inspection device includes an air blowing device, no additional device is required.

  The inspection method of the present invention can be applied to any processed material such as metal, resin, ceramic, and the like. Since the inspection method of the present invention can accurately detect defects even in a processed product having a fine unevenness, brightness difference, or color difference on the surface, it can be applied to an inspection of a metal processed product among the processed products of the above materials. Significant. Also, the processing method of the processed product is not limited, but the background and the defect can be identified even in the processed part where the step due to the screw thread is formed in a stripe shape like the screw hole. Accurate defect detection is possible even in the inspection of burrs generated during processing. The screw hole can detect a defect with a bottomed hole or a through hole. In the case of the through hole, the inspection table on which the inspection object is placed is photographed together with the inspection object. However, since the shape of the inspection table does not change in the comparison between the image during air blowing and the image without air, it is determined that there is no defect.

  The inspection result by the inspection method of the present invention can be used as a pass / fail judgment material for a processed product according to the presence or absence of burrs or the size of burrs, and further for determining whether or not burrs must be removed after inspection. Can be used.

  Below, the image | photographed image is illustrated and the inspection method of the processed goods of this invention is demonstrated.

  The object to be inspected (10) shown in FIG. 1 is an aluminum metal plate processed product comprising a flat plate portion (11), and one surface of the flat plate portion (11) is an inspection surface (11a). The entire inspection surface (11a) is subjected to hairline processing, and a plurality of bottomed screw holes (13) are formed. The screw hole (13) is formed by tapping, and there is a possibility that burrs are generated in the processed portion. Further, there is a possibility that foreign matters such as chips and dust generated during the processing are attached to the inspection surface (11a). These foreign matters are not connected to the object to be inspected (10), but are merely placed.

  The airless image and the air blow image are taken from the same direction in the same area. FIG. 1 shows the arrangement of the apparatus during image taking during air blowing. Air blown from the nozzle (5a) of the air blowing device (5) is blown onto the inspection surface (11a) of the object to be inspected (10). Shoot the area. The airless image is taken either before or after the air blown image is taken, or both. The photographed image is sent to the image analysis device (3), and a predetermined image analysis is performed to detect a defect.

  Note that the position of the nozzle (5a) and the air blowing direction shown in FIG. 1 are merely examples of these, and indicate the position of the nozzle (5a) and the air blowing direction in the following inspection examples. is not.

[Photographed image 1]
FIG. 2 shows an imaging region including one screw hole (13). This imaging region is divided into a plurality of regions, and image analysis is performed for each of the divided regions. A portion surrounded by a solid line in the image of FIG. 2 is one analysis region (A) divided. 3A is an enlarged view of the analysis region (A) in the image before air blow (30), FIG. 3B is an enlarged view of the analysis region (A) in the image during air blow (31), and FIG. 3C is an analysis region in the image after air blow (32). It is an enlarged view of (A). In each image, the position and form of the background hairline processing part (12), the screw hole (13) and its tapping part (14), and the elongated defect candidate (20) appear as gradation values of each pixel.

[Photographed image 2]
The same analysis area (A) as the photographed image 1 was photographed by changing the air blowing conditions (air blowing speed, blowing direction).

  FIG. 4A is a pre-air blow image (33), which is an image taken under the same conditions as in FIG. 3A. FIG. 4B is an air blowing image (34).

[Photograph 3]
FIG. 5 shows an imaging region including another screw hole (13). This imaging region is divided into a plurality of regions, and image analysis is performed for each of the divided regions. A portion surrounded by a solid line in the image of FIG. 5 is one analysis region (B) divided. 6A is an enlarged view of the analysis region (B) in the image before air blowing (35), FIG. 6B is an enlarged view of the analysis region (B) in the image during air blowing (36), and FIG. 6C is an analysis region in the image after air blowing (37). It is an enlarged view of (B). In each image, the background hairline processing part (12), the screw hole (13) and its tapping part (14), the elongated first defect candidate (21), and the massive second defect candidate (22) are the floors of each pixel. Its position and form appear as a key value.

[Inspection Example 1]
Inspection was performed based on the pre-air blow image (30) in FIG. 3A and the in-air blow image (31) in FIG.

  In FIG. 7, the gradation difference obtained by subtracting the gradation value of the corresponding pixel of the image during air blowing (31) from the gradation value of each pixel of the pre-air-blowing image (30) is binarized with a predetermined threshold value. The binary values are shown in shades. In the drawing, a dark color portion indicates a pixel having a difference in tone value between two images (30) and (31), and a light color portion indicates two images (30) and (31). ) Shows pixels in which there is no difference in gradation values. In other words, the dark color indicates a portion where the gradation value has changed in the two images (30) and (31), and the light color indicates a portion where the gradation value has not changed. According to FIG. 7, since the hairline process part (12), the screw hole (13), and the tap process part (14) are light colors, it has shown that it does not move even if it blows air. On the other hand, since the defect candidate (20) is dark in color, it moves when air is blown, and the defect candidate (20) is determined to be “burr”.

  Here, when the two images (30) and (31) in FIGS. 3A and 3B are observed, the defect candidate (20) appears as an elongated form in which the longitudinal direction obliquely extends from the upper right to the lower left. Then, it can be seen that the defect candidate (20) in the two images has no change in position in the upper right portion even when air is blown, and the position of the lower left portion has moved upward. Therefore, it can be determined that the defect candidate (20) is a burr that is connected to the inspection object (10) in the upper right portion and the lower left portion is blown upward by the blowing of air.

  Note that the gradation difference can be used to detect burrs without being binarized, but the process can be simplified by binarizing. Also, burrs can be detected by obtaining the difference after binarizing the gradation values of two images to be compared.

[Inspection Example 2]
In the photographed image 1, an inspection was performed based on the air blowing image (31) in FIG. 3B and the air blown image (32) in FIG. 3C.

  In FIG. 8, the gradation difference obtained by subtracting the gradation value of the corresponding pixel of the image after air blowing (32) from the gradation value of each pixel of the image during air blowing (31) is binarized with a predetermined threshold value. The binary values are shown in shades. In the drawing, the dark color portion indicates a pixel having a difference in tone value between the two images (31) and (32), and the light color portion indicates the two images (31) and (32). ) Shows pixels in which there is no difference in gradation values. In other words, the dark color indicates a portion where the gradation value has changed in the two images (31) and (32), and the light color indicates a portion where the gradation value has not changed. According to FIG. 8, since the hairline process part (12), the screw hole (13), and the tap process part (14) are light colors, it has shown that it does not move even if it blows air. On the other hand, since the defect candidate (20) is dark in color, it moves when air is blown, and the defect candidate (20) is determined to be “burr”.

  Here, when the two images (31) and (32) in FIG. 3B and FIG. 3C are observed, the defect candidate (20) appears as an elongated form whose longitudinal direction obliquely extends from the upper right to the lower left. Then, it can be seen that the defect candidate (20) in the two images has no change in the position of the upper right portion when the air blowing is stopped, and the position of the lower left portion has moved downward. Therefore, it can be determined that the defect candidate (20) is a burr that is connected to the inspection target (10) in the upper right portion and the lower left portion is blown upward by the blowing of air.

  As shown in Test Example 1 and Test Example 2, the image during air blowing (31) can detect burrs by comparing with either the pre-air blow image (30) or the post-air blow image (32). .

[Inspection Example 3]
Inspection was performed based on the pre-air blow image (33) of FIG. 4A and the in-air blow image (34) of FIG.

  In FIG. 9, the gradation difference obtained by subtracting the gradation value of the corresponding pixel of the image during air blowing (34) from the gradation value of each pixel of the pre-air-blowing image (33) is binarized with a predetermined threshold value. Values are shown in shades. In FIG. 9, the dark color indicates a portion where the gradation value has changed in the two images (33) and (34), and the light color indicates a portion where the gradation value has not changed.

  According to FIG. 9, the hairline processed part (12), the screw hole (13), and the tapped processed part (14) are light-colored as in the case of inspection examples 1 and 2, so that they do not move even when air is blown. Show. On the other hand, since the defect candidate (20) is dark in color, it moves when air is blown, and it is determined that the defect candidate is “burr”.

  Here, when the two images (33) and (34) in FIGS. 4A and 4B are observed, the defect candidate (20) appears as an elongated form in which the longitudinal direction obliquely extends from the upper right to the lower left in both images. However, the area of the portion appearing as the defect candidate (20) in the air blowing image (34) is larger than the area of the defect candidate (20) in the pre-air blowing image (33). This is because the burr vibrates due to the blowing of air, and the burr was photographed larger than the actual area by the vibration.

[Inspection Example 4]
Inspection was performed based on the pre-air blow image (35) of FIG. 6A, the in-air blow image (36) of FIG. 6B and the post-air blow image (37) of FIG.

  First, the image before air blowing (35) in FIG. 6A and the image during air blowing (36) in FIG. 6B are compared. In FIG. 10, the gradation difference obtained by subtracting the gradation value of the corresponding pixel of the image during air blowing (36) from the gradation value of each pixel of the pre-air-blowing image (35) is binarized with a predetermined threshold value. Values are shown in shades. In FIG. 10, the dark color indicates a portion where the gradation value has changed in the two images (35) and (36), and the light color indicates a portion where the gradation value has not changed.

  According to FIG. 10, since the hairline processing part (12), the screw hole (13), and the tap processing part (14) are light colors as in the inspection examples 1 to 3, they are not moved even if air is blown. Show. On the other hand, since the first defect candidate (21) and the second defect candidate (22) are dark in color, they are moved by air blowing, and the defect candidate is provisionally determined to be “burr”.

  Next, the in-air-blowing image (36) in FIG. 6B and the after-air-blowing image (37) in FIG. 6C are compared. In FIG. 11, the gradation difference obtained by subtracting the gradation value of the corresponding pixel in the post-air-blowing image (37) from the gradation value of each pixel in the air-blowing image (36) is binarized with a predetermined threshold value. Values are shown in shades. In FIG. 11, the dark color indicates a portion where the gradation value has changed in the two images (36) and (37), and the light color indicates a portion where the gradation value has not changed.

  According to FIG. 11, since the first defect candidate (21) is expressed in dark color among the two defect candidates (21) and (22) tentatively determined to be burr based on FIG. When the spraying is stopped, it is determined that the burrs exist by changing the form. On the other hand, since the second defect candidate (22) is represented in a light color, it indicates that there is no change and no burr even when the air blowing is stopped. Then, taking into account the difference between the pre-air-blowing image (35) and the air-blowing image (36) shown in FIG. 10, the change is caused by blowing air, and the change continues even if the air is stopped. It can be judged that the second defect candidate (22) is a foreign substance that is merely placed on the object to be inspected (10) such as chips and dust. Furthermore, the second defect candidate (22) is represented by a dark color in the gradation difference between the image before air blow (35) and the image after air blow (37) shown in FIG. 12, and the form is the second defect candidate in FIG. Since it is the same as (22), it can be determined that the second defect candidate (22) is a foreign matter removed by air blowing. In addition, since the first defect candidate (21) is shown in light color in FIG. 12, it confirms that the first defect candidate (21) is a burr and has returned to its original form by stopping the blowing of air. .

  In the inspection method described above, in order to reliably detect burrs from the difference in brightness due to the background, and to reliably distinguish burrs from foreign objects that are merely placed on the object to be inspected, change the form of burrs, In addition, the air blowing speed and the blowing time are set so that the foreign matter can be blown out of the imaging region. Further, since there is a possibility that foreign matter is not blown off immediately after the start of air blowing, it is preferable to take an image during air blowing immediately before the end of air blowing. Note that, depending on the connection state of the burrs, there are some that are blown off from the object to be inspected by blowing air, but such burrs are detected as foreign matters.

  In the inspection method of the present invention, the method for comparing the image during air blowing and the image without air is not limited to the above-described method for comparing the gradation values of the respective pixels, and a known image analysis method can be used as appropriate. Examples of other comparison methods include an edge extraction method and pattern matching. The edge extraction method detects a part where the data of adjacent pixels in the image changes abruptly, extracts the boundary (edge) of the change, and compares the boundary extracted from the two images to detect the shape change of the defect. It is a method of detection. As the pixel data, tone values representing light and dark and tone values representing color tone can be used as appropriate. Pattern matching is a method in which two images are overlapped to check whether or not they match, and a mismatched portion is detected as a defect.

  In the above-described inspection examples 1 to 4, the captured image is divided into a plurality of regions, and one inspection example of the divided regions is shown. By analyzing the entire divided area by the same method, burrs and foreign matters can be inspected in the entire area of the photographed image.

  Image analysis can be performed on the entire area without dividing the captured image (images in FIGS. 2 and 5), but the image is divided into small areas for image analysis as in the above inspection example. By performing the above, it is possible to distinguish a defect from noise components such as a flickering of illumination and a positional shift caused by vibration of an inspection object, and it is possible to perform an accurate and stable inspection while preventing erroneous determination. Moreover, since feature extraction is not required by analyzing with a small area, it can be inspected with a relatively inexpensive apparatus. The size of one side of the divided analysis region is preferably in the range from the minimum size of the defect to be detected to several times its size. This is because if the analysis region becomes too small, the possibility of erroneously determining a noise component as a defect increases. It is also preferable to appropriately employ a known noise component removal method such as setting the sum or average value of the gradation values of adjacent pixels as the gradation value of those pixels.

  In addition, when a defect extends over a plurality of divided areas, it is determined that there is a defect smaller than the actual in each analysis area. In such a case, more accurate determination can be performed by using a known image analysis method. For example, a method of dividing a captured image into smaller areas than the analysis area, forming an analysis area of a predetermined size while shifting the divided small areas, and comparing the air blowing image and the no air image in those analysis areas it can.

  The present invention can be suitably used for inspection of a processed product in which burrs are generated by processing.

1 ... Inspection device
2 ... CCD camera
3 ... Image analyzer
4 ... Slope illumination
5… Air blow device
5a… Nozzle
10 ... Metal plate processed product (inspected object, processed product)
11: Flat plate
11a ... Inspection surface
12 ... Hairline processing department
13 ... Screw hole
14 ... Tapping part
20, 21, 22 ... defect candidates
30, 33, 35 ... Image before air blow (image without air)
31, 34, 36 ... Air blow image
32, 37 ... Image after air blow (image without air)

Claims (6)

And air blow before the image is an air without images taken without blowing air before blowing air into the workpiece, the air blow in an image photographed while blowing air into the workpiece, the air after blowing air to the workpiece Compared with the image after air blowing, which is an image without air taken without spraying, and based on the shape change due to air blowing, it is connected to the burr and the processed product as defects that are connected to the processed product as defects. A method for inspecting a processed product, characterized by detecting a foreign object that is merely placed and not placed . The method for inspecting a processed product according to claim 1 , wherein a change in the form of a defect is detected based on a gradation difference between the image during air blowing and an image without air. In the air-blowing image and the air-free image, a portion where the data of adjacent pixels changes rapidly is detected, the boundary of the change is extracted, and the defect shape change is detected by comparing the boundary extracted from the two images. The method for inspecting a processed product according to claim 1 . The method for inspecting a processed product according to claim 1 , wherein a form change of a defect is detected by pattern matching between the air blowing image and the airless image. The workpiece inspection method of the workpiece according to any one of claims 1 to 4 having a threaded bore as a processing unit. An air blowing device that blows air onto the workpiece;
A CCD camera wherein while blowing air to the workpiece capturing images in air blow, and to shoot the air without images air blow before image and an air blow after image without blowing air after spraying and prior to blowing air into said workpiece ,
Comparing the image during air blowing and the image without air, and based on the form change caused by air blowing , the burr that is a projection connected to the processed product as a defect and the processed product are not connected An apparatus for inspecting a processed product, comprising: an image analysis device that detects only foreign matter .