JP3503130B2 - Surface inspection equipment - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a surface inspection device for inspecting the surface of an object to be inspected, for example, surface defects such as irregularities on the coating surface of an automobile body, and for classifying the surface defects.

[0002]

2. Description of the Related Art As a conventional surface inspection apparatus, for example,
There is one disclosed in Japanese Patent Application Laid-Open No. 5-322543.

The surface inspection apparatus disclosed in the above publication irradiates the surface to be inspected with bright and dark light having a light and dark gradient, and reflects the reflected light into CC.
A device for capturing defects with a D camera, detecting defects based on changes in brightness in the received light image, and determining types of defects such as unevenness based on the direction of the brightness gradient of the received light image and the image of brightness gradient of the defect portion. Is.

[0004]

By the way, in automatic coating lines for automobiles and the like, inspectors regularly carry out detection of coating defects, identification of defect types and statistical processing such as the number of defects in order to ensure stable coating quality. It takes a lot of time and time. In this case, it takes a very long time to identify the defect of the coating machine or the defect of the paint, and it is difficult to take an early countermeasure against the defect generation source at present.

The types of painting defects that occur in the current automatic painting line include general unevenness defects such as dust spots and cissing, as well as paint color spots and painting work that occur due to mixing of the previous paint when the paint color is changed. However, in the conventional surface inspection device as described above, only the general unevenness defects such as dust spots and cissing can be identified in the conventional surface inspection apparatus, and therefore all these types of defects are detected. There is a problem that it is very difficult to make an accurate determination without lowering the accuracy.

The present invention has been made in view of the above problems, and its object is not only general uneven defects such as dust spots and cissing defects but also defects such as paint color spots and scratches. At the same time, and all of the above defects can be detected without degrading the accuracy, which allows quick and accurate countermeasures against the defect source when a defect occurs. To provide an inspection device.

[0007]

A surface inspection apparatus according to claim 1 of the present invention irradiates a surface to be inspected of an object to be inspected with light,
What is claimed is: 1. A surface inspection apparatus for forming a light-receiving image based on light reflected from a surface to be inspected, and detecting defects existing on the surface to be inspected based on the light-receiving image, which is a gate-like shape surrounding an object to be inspected. Irradiating means for forming a light-dark stripe pattern having a narrow light-dark stripe pattern and a wide light-dark stripe pattern on the surface to be inspected and including a light source arranged in An image pickup unit that is fixedly arranged and forms a plurality of received light images based on the reflected light from the surface to be inspected, and an edge detection process is performed on the received light image obtained by the image pickup unit to extract only components having a predetermined brightness level or higher. Inspection processing means for detecting defects on the surface to be inspected and outputting the detection information, defect information collecting means for collecting defect information obtained by the inspection processing means, and the defect information collecting means Defect type determination means for determining the type of defects collected by, information input means for inputting the type of the object to be inspected and paint information of the surface to be inspected, defect type information determined by the defect type determination means and the Based on the defect statistical processing means for statistically processing at least the defect type and the number of defects based on the input information input by the information input means, and the statistical processing information and the empirical defect identification information obtained by the defect statistical processing means. And a source estimating means for estimating the source of the defect. In the image information having a plurality of types of light and dark pattern areas obtained by the image pickup means, the inspection processing means calculates a pitch width of the light and dark patterns and compares the pitch width with a predetermined value, and an area with a narrow pitch width or a wide area. Area determining means for determining whether
A first threshold value calculating means for calculating a first threshold value by performing a predetermined calculation based on a physical quantity relating to noise generated in the processed image information subjected to the edge detection processing; Second threshold value calculating means for calculating a second threshold value by performing a predetermined calculation based on a physical quantity related to the edge brightness level, and area image information having a wide pitch width judged by the area judging means. First binarization processing means for performing a first binarization processing using the first threshold value, and the second binarization for the area image information with a narrow pitch determined by the area determination means. Second binarization processing means for performing the second binarization processing using the threshold value of, by the first and second binarization processing,
Defects on the surface to be inspected are detected and the detection information is output.

A surface detecting device according to claim 2 of the present invention comprises:
The defect information collecting means is configured to collect the defect size information and the detection processing information obtained by the inspection processing means.

The surface inspection apparatus according to claim 3 of the present invention is
The defect type determining means is configured to determine the type of defect based on the defect size information and the detection processing information collected by the defect information collecting means.

A surface inspection apparatus according to claim 4 of the present invention is
The defect statistical processing means, based on the defect type information determined by the defect type determining means and the input information input to the information input means, for each defect type that exists for each type of coating color and each part of the surface to be inspected. The number of defects is statistically processed.

A surface detecting device according to claim 5 of the present invention is
A light diffusion sheet having a predetermined light-dark pattern for irradiating the surface to be inspected with the light from the light source of the illuminating device as surface illumination by diffused light of the light-dark pattern is provided between the light source and the object to be inspected. The light-dark pattern formed on the light diffusion sheet is a composite light-dark stripe pattern having a narrow light-dark stripe pattern and a wide light-dark stripe pattern.

The surface detecting apparatus according to claim 6 of the present invention is
A plurality of light diffusion sheets having a predetermined light-dark pattern for irradiating the surface to be inspected with the light from the light source of the illumination means as surface illumination by the diffused light of the light-dark pattern are provided between the light source and the object to be inspected. The plurality of light diffusing sheets include a light diffusing sheet having a light and dark stripe pattern having a narrow pitch width and a light diffusing sheet having a light and dark stripe pattern having a wide pitch width.

A surface inspection apparatus according to claim 7 of the present invention is
A light diffusion sheet having a predetermined light-dark pattern for irradiating the surface to be inspected with the light from the light source of the illuminating device as surface illumination by diffused light of the light-dark pattern is provided between the light source and the object to be inspected. The light-dark pattern formed on the light diffusion sheet is a continuously changing pitch-type light-dark stripe pattern in which the pitch width is continuously changed from a light-dark stripe pattern having a narrow pitch width to a light-dark stripe pattern having a wide pitch width in sequence. ing.

According to an eighth aspect of the present invention, in the surface inspection apparatus, the inspection processing means processes the inspection processing means and the moving amount of the object to be inspected from temporally different image information captured by the same imaging means. And tracking processing means for tracking and detecting image information that matches the moving direction under a predetermined condition.

[0015]

[0016]

[0017]

In the surface detecting apparatus according to claim 9 of the present invention, the area determining unit determines the pitch width of the light-dark pattern in the image information having the plurality of light-dark pattern areas and the non-inspection surface area obtained by the image capturing unit. It is calculated and compared with a plurality of predetermined values to determine whether it is a narrow pitch area, a wide pitch area, or a non-inspection surface area.

[0019]

According to the surface inspection apparatus of the first aspect of the present invention, a bright-dark pattern having a narrow bright-dark stripe pattern and a wide bright-dark stripe pattern is formed by irradiation on the surface of the object to be inspected, and the bright-dark pattern is formed. It is possible to pick up an image of the surface to be inspected in which the image is reflected and automatically and accurately detect the defect on the surface to be inspected based on the obtained light receiving image. In particular, not only general irregularities such as dust spots and craters, but also defects such as paint spots and scratches can be accurately identified at the same time, which allows the defect source to be quickly and accurately estimated when a defect occurs. Then, desired measures can be taken. Further, in the obtained image information, the defect information on the surface of the object to be inspected and the noise generated in the signal processing are clearly identified, and the defect is detected with high accuracy. Furthermore, the efficiency of the detection process when processing the obtained image information to detect a defect is achieved, and the defect is detected in a shorter time.

According to the surface inspection apparatus of the second aspect of the present invention, the defect size information such as the area and shape of the defect portion and the detection processing information by the inspection processing means are collected as the defect information. The information obtained can be used to determine the type of defect.

According to the surface inspection apparatus of the third aspect of the present invention, the type of defect is determined based on the defect size information and the detection processing information collected as the defect information.
The type of defect can be specified more quickly and with high accuracy.

According to the surface inspection apparatus of the fourth aspect of the present invention, the source of the defect can be more accurately specified by the statistical processing result of the number of defects and the empirical defect information on the production line.
It is possible to promptly feed back instructions upstream from the painting line.

According to the surface inspection apparatus of the fifth aspect of the present invention, the light and dark pattern can be accurately formed on the surface to be inspected of the object to be inspected, and the surface defect can be detected easily and highly accurately. it can.

According to the surface detecting apparatus of the sixth and seventh aspects of the present invention, not only a relatively high-angle concave-convex defect such as dust particles but also a gentle-angle concave-convex defect can be detected with high accuracy. .

With the surface inspection apparatus according to the eighth aspect of the present invention, it is possible to detect defects on the surface of the object to be inspected more quickly and with higher accuracy.

[0026]

[0027]

[0028]

According to the surface inspection apparatus of the ninth aspect of the present invention, erroneous detection can be prevented and defects can be detected with higher accuracy.

[0030]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a front view showing an embodiment of a surface inspection apparatus according to the present invention, and shows a case where a defect inspection of a painted surface of an automobile is taken as an example. As shown in FIG. 1, an arch type illumination device 1 as an illumination means formed in an arc shape so as to substantially match the cross-sectional contour of the body 5 projects a predetermined light-dark pattern on the surface to be inspected. Is configured. The light-dark pattern is a composite light-dark pattern that has a narrow light-dark pattern for detecting loose defects (low angle) and a wide light-dark pattern for detecting general uneven defects (high-angle defect). It is formed as a pattern. In addition, a plurality of CCD cameras 3 as image pickup means are fixed to a camera stand 2 that has a substantially same shape as the illumination means and is provided side by side through an adjustment fixing jig 3 ', and a bright and dark pattern is projected. Each of them is arranged at a predetermined position so as to image the surface to be inspected (see FIG. 1).
Shows one CCD camera 3. ) Furthermore, these C
The CD camera 3 is connected to the inspection processing means 4 which detects a defect on the surface to be inspected based on the light-receiving image obtained by these and outputs the result. This inspection processing means 4
Is composed of an image processing (emphasizing) unit 41, a tracking processing unit 42, and a host computer 43. In addition, as shown in FIG. 2, the image processing unit 41 performs a predetermined calculation based on the edge detection unit 13 and a physical quantity related to noise (excluding edges of the light and dark stripe pattern) generated in the image subjected to the edge detection processing. A first threshold value calculating unit 14 for calculating the threshold value of the
Binarization processing unit 16, and a second threshold value for calculating a predetermined threshold value by performing a predetermined calculation based on a physical quantity related to an edge brightness level of a stripe pattern boundary in an image subjected to edge detection processing. The calculation unit 15, the second binarization processing unit 17 that performs binarization processing based on the threshold value, calculates the width of the bright and dark stripe pattern having a high brightness level, and compares the width with a plurality of predetermined values to determine the stripe pitch width. Is a narrow area, a wide area, or a non-inspection surface area.

Further, the inspection processing means 4 detects the defect detection information obtained by the inspection processing means 4, that is, the defect size information such as area and shape, the type of binarization processing, and the number of tracking processings. The defect information collecting means 6 for collecting information is connected. Further, the defect information collecting means 6 detects all the coating defects from the defect size information such as the area and shape collected by the defect information collecting means 6 and the detection processing information such as the binarization processing type and the number of tracking processing. A defect type determination means 7 for determining the type, that is, the type of the detected defect such as dust spots, paint spots, craters, and scratches is connected.

Further, the defect type determining means 7 has information on the type of the object to be inspected, that is, the type of vehicle and the paint on the surface to be inspected, input by the information inputting means 8 and the defect type determining means 7.
Based on the defect type information determined by, the defect statistical processing means 9 for statistically processing the type of coating color and the number of defects for each defect type existing for each part of the surface to be inspected is connected.

Further, the defect statistical processing means 9 specifies the source of the defect based on the statistical processing information obtained by the defect number statistical processing means 9 and the empirical defect defect specifying information on the coating line. The source estimating means 10 is connected, and the identification result is output by the display / instructing means 11 such as a printer.

The processing flow in the surface inspection apparatus having the above structure, as shown in FIG. 3, first, the defect on the inspected surface based at step S _1, the inspection processing unit 4 to the received-light image of the CCD camera 3 To detect. Subsequently, in step S _2, the defect information acquisition means 6 for collecting defect detection information obtained by the inspection processing unit 4. further,
In step S ₃ determines, collected types of defects repelling, coating lumps, Gomibutsu, which one type of scratches by defect type determination unit 7. Next, in step S _4, enter the paint information models and test surface using an information input unit 8, based on the determined defect type information and the input information, in step S _5, the defect statistical processing unit 9 Thus, the number of defects for each defect type, and further the number of defects for each defect type existing for each type of coating color and inspection surface are statistically processed. Subsequently, in step S _6, based on empirical defect specific information in the statistical processing information and painting line, to identify the source of a defect by source estimation means 10. Finally, display / instruction means 11 such as a printer
Displays the specific information of the above-mentioned generation source and announces it.

Here, a method of forming a bright / dark pattern by the illumination means 1 and the illumination means will be described.

The illuminating means 1 has a back plate which is white or surface-treated so as to diffuse and reflect the light on the back side of the light source in order to irradiate the light from the light source onto the surface to be inspected without waste and without unevenness. It has a structure in which a plurality of light sources are attached to a face plate at substantially equal intervals.

Further, U-shaped fluorescent lamps are attached to the back plate at substantially equal intervals. For example, four fluorescent lamps in two rows are attached to one back plate on the side of the object to be inspected.
A power supply for turning on the fluorescent lamp at a high frequency is attached to the back side thereof to form an illumination unit. The above-mentioned lighting unit 1 can be realized by mounting such a lighting unit on an arch-shaped pillar (not shown) without a gap.

FIG. 4 is a perspective view showing the structure of the light-dark pattern sheet 105 and the sheet guide 106 as a light diffusion sheet. The light-dark pattern sheet 105 has a function of diffusing light and is of an object to be inspected. For example, a sheet-like material which can be easily deformed so as to follow the contour shape of the cross section and which is provided with a predetermined light-dark pattern such as matte black masking tape. The reason for diffusing the light is to suppress the influence of the metallic glittering material when the body 5 is metallically painted. Furthermore, the light and dark pattern sheet (diffusion sheet) 10
No. 5 is formed as a composite type bright / dark stripe pattern having a narrow bright / dark stripe pattern and a wide bright / dark stripe pattern in combination.

If the diffusing sheet 105 is made of a material that easily causes wrinkles and the wrinkles cause shadows or illumination spots, the diffusing sheet 105 is supported from below as shown in FIG. By using the sheet guide 106 for stretching the cross-sectional contour shape, it is possible to suppress the occurrence of wrinkles. The sheet guide 106 is painted matt black, and the columns of the sheet guide are spaced so as to overlap the dark portion of the light-dark pattern of the diffusion sheet 105. Diffusion sheet 105 and sheet guide 106
And are fixed to the column using matte black bolts, nuts, washers, etc. (not shown).

Further, as shown in FIG. 4, casters 106 'are attached to the sheet guide 106 so that the sheet can be moved with the diffusion sheet 105 stretched.

Another embodiment of the light-dark pattern is shown in FIG.
As shown in, the light diffusing sheet 205 that constitutes a continuously changing pitch type light-dark stripe pattern that continuously changes from a light-dark stripe pattern having a narrow pitch width to a light-dark stripe pattern having a wide pitch width can be adopted.

Next, the inspection processing means for detecting a defect on the surface to be inspected based on the received light image obtained by the image pickup means and outputting the result will be described. 6 and 7 show an example of image processing and an example of a processing flow in the inspection processing means 4 for detecting a general uneven defect (relatively high angle) on the coated surface.

When the CCD camera 3 captures an image of the surface to be inspected on which the bright and dark patterns are projected by the illumination means, FIG.
It becomes like (a). In this original image, light is diffusely reflected at the concave and convex defect portions, so that the bright pattern 1
At 41, it appears as a dark area 143. Similarly, when the defect is in the dark pattern 142, it appears as a bright portion.

When this original image is subjected to edge detection processing such as differentiation and binarized with a predetermined threshold value, FIG.
As shown in FIG. 5, the area where the brightness changes in the image, that is, the area 144 with high spatial frequency is white, and the other part 1
A binary image in which 46 is black is obtained.

Next, labeling (numbering) and area / center of gravity calculation are performed on the white pixels of this binary image. The defect appears as an isolated point in the white pixel of the binary image, and the boundary line 144 of the light-dark pattern becomes a large object that crosses the top and bottom of the screen. Therefore, the area determination is performed by a predetermined determination value and only the isolated point having a small area is detected. Is extracted, an image as shown in FIG.

Here, if there is unevenness on the coated surface that does not become a defect such as orange peel, it is extracted as an isolated point 145 (hereinafter, referred to as noise) together with the defect 143, as shown in FIG. 6C. There is a case.

Further, FIG. 8 shows an image processing (enhancement) for detecting a general uneven defect (relatively high angle) on the coated surface shown in FIGS. 6 and 7 as well as a loose uneven defect. ) Shows an example of a flow.

Here, the detection of a general uneven defect (defect having a relatively high angle) is the same as that shown in FIGS. When the CCD camera 3 captures an image of the surface to be inspected on which the bright and dark patterns are projected by the illumination means, the result is as shown in FIG. In this original image, the irregular irregular reflection (angle) of light is small in the loose irregular defect portion, so it is difficult to detect the loose irregular defect portion at the center of the bright-dark pattern.

On the other hand, in the light-dark pattern boundary portion as shown in FIG. 9A, the stripe pattern on the opposite side can be reflected due to the positional relationship of the light-dark stripes, and therefore, on the edge processing screen such as differentiation, the boundary pattern is on this boundary. The defect has a high luminance point as shown in FIG. When the binarization processing is performed on this differential screen using the threshold value of the brightness average + α of the bright / dark pattern boundary portion (stripe boundary) obtained by the calculation (FIG. 9).
(E)), as shown in FIG. 9 (c), a binary image in which a region with a large change in luminance, that is, a region (defect) with a high spatial frequency is white, and the other part including the bright / dark pattern boundary is black is can get. Labeling (numbering) and area / center of gravity calculation are performed on the white pixels of this binary image. Since a defect is an isolated point in a white pixel of a binary image, if an area determination is performed with a predetermined determination value and only an isolated point having a small area is extracted, an image as shown in FIG. 9D is obtained.

Here, an example of the binarization processing by the inspection processing means shown in FIGS. 8 and 9 will be described.

As shown in FIG. 10 and FIG. 11, the bright / dark stripe original image is divided by the area determination means into an area having a small pitch width (small pitch) and a wide area (large pitch) in the bright portion or the dark portion. It For this narrow area, the second threshold value (threshold value 2) calculated by the second threshold value calculation means is used based on the physical quantity relating to the edge brightness level in the processed image information subjected to the edge detection processing. Then, the second binarization processing is performed. In addition, for a wide area, the first threshold value (threshold value 1) calculated by the first threshold value calculation means is set based on the physical quantity related to the noise generated in the processed image information subjected to the edge detection processing. It is used to perform the first binarization process. As a result, the efficiency of the detection process is achieved. When the original image of the light and dark stripes has a non-inspection surface such as a background, as shown in FIG. 12, the pitch width L of the light and dark stripes is a predetermined value L0.
In the above cases, it is determined that the area is an unnecessary area and a predetermined threshold value 3 is applied. When it is less than the predetermined value L1, it is determined that the area is a narrow area and the threshold value 2 calculated above is applied. , L0> L
In the case of> L1, it is determined that the area is wide and the threshold value 1 calculated above is applied to perform binarization processing. This can prevent erroneous detection.

Here, as the X-axis coordinate X ₀ for dividing the bright-dark stripe pattern into a region having a narrow pitch width and a region having a wide pitch width, pitches at n Y-axis coordinates Y ₁ , Y ₂ , ... Coordinates X on n X-axis with width L equal to or greater than L _{0
10, X} 20, seeking ... Xn _0, applying the average value _{X 0} obtained by substituting the following equation (1).

N Xo = Σ Xmo / n (1) m = 1 A tracking process 42 for extracting only defects from the process data obtained by the above will be described with reference to FIGS. 7 and 13. In the inspection processing means as described above, when the processing for extracting isolated points is continuously performed in time, the area determination images obtained as a result are as shown in FIGS. 13A to 13F, and these are overlapped. Then, the result is as shown in FIG.

That is, since the camera and the illumination means are fixed, while the body 5 which is the object to be inspected moves, the defect 211 on the body surface in the camera image.
Moves in the direction of the arrow in FIG. 13 (g) according to the movement of the body, but the noise 212 occurs randomly regardless of the movement of the body 5.

Therefore, it is possible to finally determine the defect 211 as the one that matches the movement amount and the movement direction of the body 5 under the predetermined condition from the continuous area determination images that are different in time.
Since the moving direction of the defect 211 in the image is determined by the direction in which the body 5 passes with respect to the camera 3, the position of the camera 3 is fixed and the carrying direction of the body 5 as in the present embodiment. Can be determined for each camera if is always the same.

Further, if the field of view of the camera 3 is set parallel to the carrying direction of the body 5, the defect will move in the horizontal or vertical direction in the image. In this embodiment, as shown in FIG. 13G, the camera 3 is assumed to be fixed in a direction in which the defect 211 moves right in the image (arrow).

In the thus obtained temporally different images, first, the barycentric coordinates in the Y direction (vertical direction of the screen) of the white pixels of each image are compared. As described above, since the defect 211 moves right sideways in the image, if there is a white pixel having substantially the same barycentric coordinates in the Y direction between the two images, it is determined that the white pixel is likely to be the defect 211. it can. Therefore, the white pixel is stored in the memory as a defect candidate.

Next, in the white pixel in the defect candidate,
The difference in the barycentric coordinates in the X direction between the two images represents the number of moving pixels in the images, and the arrangement direction thereof represents the moving direction. Therefore, the actual number of moving pixels calculated from these and the amount of body movement, and the body 5 in the image. If the difference is within a predetermined range (within a range considered to match) with the moving direction, it can be determined that the white pixel is more likely to be defective. Therefore, the temporally new X and Y barycentric coordinates of the white pixel are stored in the memory.

By repeating the above series of processing,
As a result of the above comparison, when one white pixel is matched more than a predetermined number of times, the white pixel is determined to be a defect (defect determination in FIG. 7), and the final barycentric coordinates and area are added to the defect list. Write and store.

The tracking processing means 42 continuously performs the above-mentioned processing while the body 5 is in the field of view of the camera, and after the body 5 passes, sends the defect list to the host computer 43.

Next, the defect type determination processing by the defect type determination means 7 will be described.

FIG. 14 is a flow chart showing the defect type determination process. First, various defect detection information obtained by the inspection processing unit 4, that is, defect size information such as area and shape, and whether it is the binarization process 1 or the binarization process 2, that is, the type of binarization process and The defect information collecting means 6 collects detection processing information such as the number of times of tracking processing.

Then, based on the defect size information such as area and shape collected by the defect information collecting means 6 and the detection processing information such as the binarization processing species and the number of tracking processings, as shown in FIG. , All types of coating defects such as paint color spots, craters, and scratches are judged.

First, when the binarization processing type is the binarization processing 2 and the number of times N of the tracking processing is smaller than the predetermined value N _0, it is determined that the pit is a loose concave defect. Similarly, when the binarization processing type is the binarization processing 2 and the detection level frequency is high, that is, when the number N of times of tracking processing is a value of N ₀ or more, it is determined that the paint color is solid.

On the other hand, when the binarization processing type is the binarization processing 1, it is judged whether the size ratio γ is larger or smaller than the predetermined value γ ₀ , and when it is γ ₀ or more, the area change ratio η is further predetermined. Value η
_It is judged whether it is smaller or larger than ₀ , and if it is η ₀ or more, it is judged as a paint color spot, and if it is smaller than η ₀ , it is judged as a dust spot which is a convex defect. Further, when the size ratio γ is smaller than γ ₀ , it is determined that the scratch is a linear scratch containing dust and the like.

Next, the statistical processing in the defect statistical processing means 9 will be described. In this statistical processing, the vehicle type and the surface to be inspected are determined based on the defect type information obtained by the defect type determining means 7 and the input information such as the vehicle type, the paint type and the paint color input by the information inputting means 8. The detected defect types (dust spots, craters, paint color spots, and scratches) are classified according to the body part located, paint type, and paint color, and the number of defects for each of these defect types is statistically calculated for each predetermined reference inspection unit. Processing.

Table 1 is an example in which statistical processing is performed for each vehicle type, and Table 2 is an example in which statistical processing is performed for each paint type. Specifically, the surface of the body 5 to be inspected is divided into a vertical surface and a horizontal surface.
The upper, middle, and lower areas are divided into three areas, and the horizontal area is divided into two areas, the right area and the left area, so that the defects detected in these areas are classified according to their types, and also the defects of each defect type are classified. The number is counted and expressed as the number of defects per unit. For example, in Table 1, for vehicle type A, 1.2 dust spots / in the vertical upper region of the body 5 /
0.1 pieces of dust spots in the central area of the vertical surface of the stand and body 5
The result is that there are no garbage spots on the platform and other areas.

[0069]

[Table 1]

[0070]

[Table 2]

Next, the process of estimating the source of the defect in the source estimating means 10 will be described. FIG. 15 is a flowchart showing the defect source estimation processing. In this source estimation processing, statistical processing information obtained by the defect statistical processing means 9, specifically, the number of defects in the vehicle type classification,
The estimation process is performed based on the number of defects in the paint / color classification, the number of defects in the body part classification, the number of defects in the defect type classification, and the like. As shown in FIG. 15, first, in step S ₁ , it is determined whether or not the number of defects in the vehicle type classification is less than or equal to a predetermined value a, and if it is less than or equal to the predetermined value a, the process proceeds to step S ₂ to paint / color. It is determined whether the number of defects in the classification is less than or equal to the predetermined value b. In this determination, if the number of defects is less than the predetermined value b, the process proceeds to step S _3, the number of defects in the vehicle part classification whether less than a predetermined value c is determined. Here, if the number of defects is less than or equal to the predetermined value c, further proceeds to step S _4, it is determined the number of defects in the defect type classification is whether more than a predetermined value d, the predetermined value d is number of defects in this determination In the following cases, there is no problem on the painting line and it is judged to be normal.

On the other hand, in each of steps S _{1 to} S ₄ , the number of defects is the predetermined value a, b,
If it is determined that the difference is larger than c and d, step S ₅
Proceed to, is performed classification, such as size / shape / frequency of occurrence of defects, followed by, in step S _6, the defect defect identification information empirically obtained coating line and the above defect statistical processing information is collated It Then, in step S ₇ ,
Identifying the source of defects in the coating line configuration, in step S _8, and displays / instructs the abnormal state.

According to the above configuration, it is possible to promptly and accurately instruct countermeasures for the defect generation source when a defect occurs due to the accurate defect identification power and high detection accuracy.

[Brief description of drawings]

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

FIG. 2 is a block diagram showing an image processing unit forming a part of the surface inspection apparatus according to the present invention.

FIG. 3 is a flowchart showing a processing flow of the surface inspection apparatus according to the present invention.

FIG. 4 is a schematic perspective view showing a composite type light-dark pattern light diffusion sheet and a sheet guide which constitute a part of the surface inspection apparatus according to the present invention.

FIG. 5 is a schematic perspective view showing a continuously variable pitch type light-dark pattern light diffusion sheet and a sheet guide which constitute a part of the surface inspection apparatus according to the present invention.

FIG. 6 is a diagram showing an example of image processing by inspection processing means according to the present invention.

FIG. 7 is a flowchart showing a surface inspection process according to the present invention.

FIG. 8 is a flow chart showing another embodiment of the surface inspection processing according to the present invention.

FIG. 9 is a diagram showing another embodiment of image processing by the inspection processing means according to the present invention.

FIG. 10 is a diagram for explaining binarization processing according to the present invention.

FIG. 11 is a diagram for explaining binarization processing according to the present invention.

FIG. 12 is a diagram for explaining the binarization processing when a non-inspection surface such as a background exists in the binarization processing according to the present invention.

FIG. 13 is a diagram showing an image subjected to area determination by inspection processing means according to the present invention.

FIG. 14 is a flowchart showing a determination process by the defect type determination means according to the present invention.

FIG. 15 is a flowchart showing a process of estimating the source of a defect by the source estimating means according to the present invention.

[Explanation of symbols]

1 Lighting Means 2 Camera Stand 3 CCD Camera (Imaging Means) 4 Inspection Processing Means 5 Body (Inspected Object) 6 Defect Information Collecting Means 7 Defect Type Judging Means 8 Information Input Means 9 Defect Statistical Processing Means 10 Source Estimating Means 11 Display / Instructing means 13 Edge detection section 14, 15 Threshold value calculation section 16, 17 Binarization processing section 18 Area determination section 41 Image processing section 42 Tracking processing section 43 Host computer 105 Composite type bright / dark stripe pattern light diffusion sheet 106 Sheet guide 205 Continuously changing pitch type light and dark stripe pattern light diffusion sheet

─────────────────────────────────────────────────── ─── Continuation of front page (56) Reference JP-A-9-184713 (JP, A) JP-A-6-148090 (JP, A) JP-A-8-285559 (JP, A) JP-A-8- 50011 (JP, A) JP 9-210922 (JP, A) JP 4-204359 (JP, A) JP 2-73139 (JP, A) JP 10-318938 (JP, A) JP-A-10-9837 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G01B 11/00-11/30 102 G01N 21/84-21/958 G06T 1/00-9 / 40

Claims (9)

(57) [Claims] 1. A surface to be inspected of an object to be inspected is irradiated with light, a light receiving image is formed based on light reflected from the surface to be inspected, and defects existing on the surface to be inspected are detected based on the light receiving image. A surface inspection apparatus that includes a light source arranged in a gate shape surrounding an object to be inspected, and has a light and dark stripe pattern having a narrow light and dark stripe pattern and a wide light and dark stripe pattern on the surface to be inspected. An irradiation means for forming an image, an imaging means arranged and fixed in a gate shape surrounding an object to be inspected, and forming a plurality of light-receiving images based on the reflected light from the surface to be inspected, An inspection processing unit that performs edge detection processing on an image, detects only a component having a predetermined brightness level or higher to detect a defect on a surface to be inspected, and outputs the detection information, and defect information obtained by the inspection processing unit. Defect information collecting means for collecting, defect type judging means for judging the kind of defects collected by the defect information collecting means, information input means for inputting the kind of the object to be inspected and the paint information of the surface to be inspected, A defect statistical processing unit that statistically processes at least the defect type and the number of defects based on the defect type information determined by the defect type determination unit and the input information input by the information input unit, and the statistics obtained by the defect statistical processing unit. A source estimating means for estimating a source of a defect based on processing information and empirical defect identification information; and, wherein the inspection processing means includes a plurality of types of bright and dark pattern areas obtained by the imaging means. Area determination means for calculating the pitch width of the light-dark pattern in the image information and comparing it with a predetermined value to determine whether the pitch width is narrow or wide. First calculating a first threshold performs predetermined calculation on the basis of the physical quantity related to noise generated in the edge detection process described processing image information
And a second threshold value calculating means for calculating a second threshold value by performing a predetermined calculation based on a physical quantity relating to an edge brightness level in the processed image information subjected to the edge detection processing. First binarization processing means for performing first binarization processing using the first threshold value on area image information having a wide pitch width determined by the area determination means; and the area determination means. A second binarization processing unit that performs a second binarization process using the second threshold value on the area image information having a narrow pitch determined in Step 1. And a surface inspection apparatus that detects a defect on the surface to be inspected by the second binarization process and outputs the detection information. 2. The surface inspection apparatus according to claim 1, wherein the defect information collecting means collects defect size information and detection processing information obtained by the inspection processing means. 3. The surface inspection according to claim 2, wherein the defect type determination unit determines the type of defect based on the defect size information and the detection processing information collected by the defect information collection unit. apparatus. 4. The defect statistical processing means, based on the defect type information determined by the defect type determination means and the input information input by the information input means, for each type of coating color and each part of the surface to be inspected. The surface inspection apparatus according to claim 1, wherein the number of defects for each existing defect type is statistically processed. 5. A light diffusion sheet having a predetermined light-dark pattern for irradiating the surface to be inspected with light from the light source of the illumination means as surface illumination by diffused light in the light-dark pattern is provided between the light source and the object to be inspected. 2. The surface according to claim 1, wherein the light-dark pattern formed on the light-diffusing sheet is a composite light-dark stripe pattern having a narrow light-dark stripe pattern and a wide light-dark stripe pattern. Inspection device. 6. A plurality of light diffusion sheets having a predetermined light-dark pattern for irradiating the surface to be inspected with light from the light source of the illuminating means as surface illumination by diffused light of the light-dark pattern, the light source and the object to be inspected. The light diffusing sheets provided between the light diffusing sheets include a light diffusing sheet having a light and dark stripe pattern having a narrow pitch width and a light diffusing sheet having a light and dark stripe pattern having a wide pitch width. 1. The surface inspection device according to 1. 7. A light diffusion sheet having a predetermined light-dark pattern for irradiating the surface to be inspected with light from the light source of the illuminating means as surface illumination by diffused light in the light-dark pattern is provided between the light source and the object to be inspected. The light-dark pattern formed on the light diffusion sheet is a continuously changing pitch-type light-dark stripe pattern in which the pitch width is continuously changed from a light-dark stripe pattern having a narrow pitch width to a light-dark stripe pattern having a wide pitch width in order. The surface inspection apparatus according to claim 1, wherein 8. The inspection processing means matches the movement amount and movement direction of the inspected object under predetermined conditions based on temporally different image information processed by the inspection processing means and imaged by the same imaging means. The surface inspection apparatus according to claim 1, further comprising a tracking processing unit that tracks and detects image information. 9. The area determination means calculates the pitch width of the light and dark pattern in image information having a plurality of light and dark pattern areas and a non-inspection surface area obtained by the image pickup means, and compares it with a plurality of predetermined values. The surface inspection apparatus according to claim 1, wherein it is determined whether the area has a narrow pitch width, a wide area, or a non-inspection surface area.