Surface defect inspection apparatus

The present invention comprises an illuminating unit configured by combining a plurality of layout patterns in which light emitting elements are continuously arranged so as to leave a dark surface having a predetermined shape inside, and an inspected surface illuminated by the irradiation light of the illuminating unit. The present invention relates to a surface defect inspection apparatus including an image pickup camera that captures an image of the image, and a defect evaluation means that evaluates an output signal of the image pickup camera and detects a defect on the surface to be inspected.

A typical example of this type of inspection device is a technique used for inspecting a painted surface of an automobile body. In such a surface inspection, irregularities and scratches existing on the painted surface as the surface to be inspected are subject to the inspection. As an inspection technique using patterned inspection light, the painted surface is irradiated with illumination light that creates so-called striped, that is, vertical striped light and darkness, and the painted surface in the irradiated state is imaged by an imaging camera. There is a technique for performing a surface inspection using the obtained captured image. (Technology disclosed in Patent Document 1 and Patent Document 2).

For example, when the painted surface is moved in a predetermined direction (for example, the X direction), the image portion of the defect such as the uneven surface on the painted surface has the coordinates in the direction orthogonal to the moving direction (for example, the Y direction). Defects are detected by using the fact that images are taken while changing the directional coordinates (X coordinates) without changing them. In the captured image of the defect area, the bright stripes are dark and dark stripes. Since the defect is identified by utilizing the fact that the portion is brightly imaged, the defect is captured as a half-coordinate image of the bright portion and the dark portion of the stripe (see Patent Document 1).

In order to detect defects called "Yuzu skin", which are periodic irregularities on the surface, we try to find a spot of coating thickness by fluctuations on the captured image of the boundary line of light and dark stripes, which is the inspection light. There is one (see Patent Document 2). In this inspection method, it is not necessary to move the surface to be inspected, but in general, the painted surface is distorted so as to cause the border image of the stripe to be misaligned over a relatively wide area of the painted surface. It becomes a detection target.<patcit num="1"><text>Japanese Unexamined Patent Publication No. 8-145906 (Fig. 5, Fig. 9 and Fig. 15)</text></patcit><patcit num="2"><text>Japanese Unexamined Patent Publication No. 9-126744 (Fig. 13)</text></patcit>

<p> In the above-mentioned conventional surface inspection methods such as Patent Documents 1 and 2, since the illumination unit that illuminates the surface to be inspected irradiates the painted surface with a striped light-dark pattern, the wraparound of the irradiation light used for the surface inspection is a stripe. Since it occurs only in the direction orthogonal (crossing) to the light-dark pattern of the shape, when a defect exists on the surface to be inspected, in order to cause the wraparound of the irradiation light from many directions, the illuminating part is used. It is preferable to combine a plurality of layout patterns in which light emitting elements are continuously arranged so as to leave a dark surface having a predetermined shape inside, but a large amount of light is emitted from the image of the surface to be inspected obtained through the imaging camera. There will be a light emitting image of the element, and at that time, depending on the shape of the surface to be inspected and other conditions, the light emitting image of the light emitting elements arranged continuously will be intermittent, making it difficult to distinguish from defects. Problems arise.</p><p> In view of the above circumstances, the subject of the present invention is that even when a lighting unit configured by combining a plurality of layout patterns in which light emitting elements are continuously arranged so as to leave a dark surface having a predetermined shape inside is used, defects are found. It is to provide a surface inspection apparatus which suppresses false detection as much as possible.</p>

<p> In order to solve the above problems, the surface inspection apparatus according to the present invention includes an illumination unit configured by combining a plurality of layout patterns in which light emitting elements are continuously arranged so as to leave a dark surface having a predetermined shape inside, and the illumination. The defect evaluation means is composed of an image pickup camera that captures an image of the surface to be inspected illuminated by the irradiation light of the unit and a defect evaluation means that evaluates an output signal of the image pickup camera to detect a defect on the surface to be inspected. An isolated point extraction unit that determines an isolated protruding luminance region in the bright / dark image of the surface to be inspected as a defect candidate generated from the output signal, and a region showing a light emitting image of the continuously arranged light emitting elements in the bright / dark image. It is provided with a defect candidate selection unit that excludes the defect candidates included in the above from the defect candidates.</p><p> In this configuration, for defects existing inside the irradiation points of the light emitting elements group arranged continuously in a ring shape, that is, on the surface to be inspected facing the dark surface, the irradiation light is emitted from the entire circumference direction of the defects. A part of the defect image will be hit, and the defect image will be brightly floated in the dark dark surface image, and it will be possible to detect the defect candidate as an isolated projected brightness region in the bright and dark image, and it will be arranged continuously. In contrast to the fact that the emission image of the light emitting element is detected as an isolated protruding luminance region (also referred to as an isolated point) at the intermittent portion, an isolated point existing on an extension of the continuous emission image in a predetermined pattern is used. By excluding from defect candidates, false detection of defects is reduced. By moving the surface to be inspected relative to the imaging camera or the illumination unit, the defect always deviates from the irradiation point of the light emitting element group and comes to a position facing the dark surface, so that the protrusion brightness detected in a predetermined pattern The region of the continuous emission image as the region may be a region not subject to defect determination.</p><p> Since a lighting unit in which many light emitting elements are continuously arranged is used, the reflected light of the irradiation light of the light emitting elements on the surface to be inspected is captured by the image pickup camera, and the light emission image is captured in the image output from the image pickup camera. However, the brightness value of this luminescent image will vary depending on the inspection conditions, especially the condition of the surface to be inspected. In the present invention, since the luminance value of this emitted light image serves as an important reference for defect determination, it is a normal reference when generating the bright and dark image from the output signal from the imaging camera in order to compensate for such fluctuation. A pre-processing unit that performs image processing so that the brightness level of the light emitting image of the continuously arranged light emitting elements obtained from the surface to be inspected substantially matches the brightness level of the continuous light emitting image region at the time of actual inspection is provided. It is convenient to be done. Performing the luminance adjustment of the acquired captured image so as to match the luminance level of the luminescent image obtained with respect to the reference surface to be inspected in advance leads to the improvement of the accuracy of the subsequent defect determination.</p><p> When the surface to be inspected is smaller than the shooting field of the imaging camera, a subject (background, etc.) other than the surface to be inspected will enter the acquired image, but the position information of such an unnecessary image area can be predicted in advance. In addition, since it can be grasped by a well-known background selection algorithm, in the present invention, an unnecessary image area such as a background is added to the peripheral area including the protruding brightness area excluded from the defect candidates described above. It has also been proposed to mask the income image with the integrated area as the area not subject to defect judgment. Other features and advantages according to the present invention will be clarified by the description of the embodiments with reference to the following drawings.</p>

As an example of the surface defect inspection device according to the present invention, FIG. 1 shows a schematic configuration diagram of a device for inspecting the painted surface of the automobile body 1 after the painting process, which is conveyed to the left side of the paper surface by the conveyor 2. ing. This surface defect inspection device includes an illumination unit 3 that irradiates an illumination light as inspection light on a painted surface of an automobile body 1 that is an inspection surface, and an imaging camera 4 that captures an image of the inspection surface illuminated by the illumination unit 3. A controller 5 that evaluates the existence of defects on the surface to be inspected using the output signal from the image pickup camera 4 and outputs the evaluation defects, and a monitor as an output device connected to the output unit 10 of the controller 5. It consists of 12 and printer 13. The controller 5 has an illumination / imaging control unit 9 that controls the lighting unit 3, and an image input unit 7 that takes in the output signal from the imaging camera 4 and expands it into the memory 8 as digital image data (hereinafter simply referred to as an input image). , Defect evaluation means 6 for performing defect evaluation using an input image is provided. Further, the controller 5 is connected to the host computer 14 as a higher-level controller of the surface defect inspection device via the communication unit 11 so as to be able to transmit data. The host computer 14 stores information on the automobile body 1 to be inspected and operation information on the conveyor 2 that are downloaded to the controller 5 as needed, and further, defect information on the painted surface generated by the controller 5. Is also uploaded from controller 5 to host computer 14 and stored there. In addition, a projector 15 or a printer controlled by a terminal connected to the host computer 14 via a network is provided in the inspection verification station, and is based on defect information sent from the controller 5 of the surface defect inspection device via the host computer 14. It is configured to instruct the inspector of the defect position and the like.

The light emitting surface 3a of the illumination unit 3 and the lens surface 4a of the imaging camera 4 are arranged so as to face the surface to be inspected of the automobile body 1 conveyed by the conveyor 2, or if necessary, the light emitting surface 3a and the lens surface 4a. Follow-up control is performed with respect to the surface to be inspected so that the vertical line of the lens surface 4a and the vertical line of the surface to be inspected are as coincident or parallel as possible.

The illumination unit 3 is referred to as a large number of light emitting elements (hereinafter, since LED elements are used in this embodiment, they will be referred to as LED elements, but of course, the light emitting elements of the present invention are not limited to LED elements, and other light emitting elements. 30 is a net-like (ring-shaped) layout pattern that leaves a hexagonal space, and is continuously (between the adjacent LED elements 30) so as to repeat this hexagonal layout pattern. It has an arranged configuration (while closing). The space left by the LED elements 30 arranged in a hexagonal net is called a dark surface 31 here, and is a black or dark plate surface.

Many dark surfaces 31 are exposed by the LED elements 30 arranged in a mesh pattern, and the imaging camera 4 is illuminated so that the lens surface 4a of the imaging camera 4 is located on the dark surface 31 located at the center of the dark surface 31. It is incorporated in Part 3. The number of image pickup cameras 4 installed is appropriately determined by the size of the light emitting surface 3a of the illumination unit 3.

The controller 5 has a CPU as a core member, and a functional unit for performing various operations of this surface defect inspection device is constructed by hardware, software, or both, as shown in FIG. As a functional unit particularly related to the present invention, a preprocessing unit 60A that converts an input image expanded in the memory 8 into a form suitable for defect detection, and a defect on the surface to be inspected using the preprocessed input image. Can be divided into defect determination unit 60B to find out.

Preprocessing unit 60A includes a luminance adjustment unit 61 and the luminance adjusted 2 input image Nekasho performing brightness adjustment for the input image consists of physical binarizing unit 62. The brightness adjustment unit 61 of this embodiment is not only a gamma adjustment, but also an LED element obtained from a normal surface to be inspected in which the brightness level of the emission image included in the input image is a reference for each coating color and coating surface. The brightness of each pixel area is also adjusted so as to reach the brightness level of the emitted image. Further, the binarization processing unit 62 applies a smoothing filter to the input image to eliminate the noise and the binarization threshold determination unit 62a that determines the binarization threshold by a statistical method from the grayscale histogram of the input image. It also has an image feature extraction unit 62b that applies an edge enhancement filter such as a Sobel filter to emphasize the contours of the luminescent image and defect image, and uses the binarization threshold determined by the binarization histogram determination unit 62a to image features. The input image emphasized by the extraction unit 62b is converted into a binarized image.

An example of the input image binarized by the binarization processing unit 62 is shown in FIG. In this binarized light-dark image, the high-brightness area is displayed in white, but the LED element group, which is a light emitting image continuously arranged in a hexagonal layout pattern, is continuously connected in a hexagonal shape. The painted surface area facing the dark surface 31 is displayed as a dark area, and in some cases, existing paint defects are displayed as white independent areas floating in the dark area due to diffused reflection by the irradiation light from the surroundings. Will be done. From this, in the defect detection, in the binarized image, it is sufficient to find a region where the brightness is prominent (a white region in this embodiment) and which is not continuous in a predetermined pattern, that is, an isolated point. Become. A well-known image processing algorithm itself that searches for continuous pixels or an isolated area while having a predetermined level of brightness value (density value) can be used.

However, the surface to be inspected is interrupted by the light emission image of the LED element 30, which originally appears as a continuously connected line, as shown in an enlarged view in FIG. May occur, and the interrupted part may be erroneously detected as a defect. The defect determination unit 60B is substantially composed of a program so as to appropriately avoid such false positives. That is, the defect determination unit 60B is a defect candidate extraction unit 63 that detects a discontinuous independent pixel area composed of a predetermined number of pixels or less as an isolated point and sets it as a defect candidate, and a continuously arranged LED element. The defect candidate selection unit 64, which excludes defect candidates included in the region showing the 30 emission images from the defect candidates, and the isolated point region and the unnecessary image area such as the background excluded from the defect candidates by the defect candidate selection unit 64 are integrated. Image mask generation unit 65 that performs mask processing as a defect judgment target area, and labeling processing that assigns different labels (numbers) to different defect candidate areas in order to identify a plurality of defect candidate areas located outside the image mask. The label setting unit 66 to be performed, the area calculation unit 67 that calculates the area of each labeled defect candidate area, and the defect candidate is determined as a true defect based on the area information from the area calculation unit 67 and used as a defect map. It is provided with a defect determination unit 68 for writing. The defect candidate selection unit 64 checks whether or not the defect candidates are extracted as defect candidates a predetermined number of times from the images sequentially sent from the image pickup camera 4 in order to select the defect candidates extracted by the defect candidate extraction unit 63. A light emitting image in which the defect candidate time series determination unit 64a, which prevents the sudden bright region from being recognized as a defect candidate, and the defect candidates (isolated points) extracted from FIG. 5 are continuous. It is equipped with a light emitting image discontinuous part search unit 64b that prevents the interrupted part of the light emitting image from being recognized as a defect candidate by checking whether or not it is located on an extension line. This search for the non-continuous part of the luminescent image is performed by using a shape feature extraction algorithm or the like that extracts a dark region located in an extension line region of the break while tracing continuous luminescent image pixels.

The procedure for evaluating defects on the painted surface by the defect evaluation means 6 configured in this way will be described below using the flowchart of FIG. First, the frame images sequentially sent from the image pickup camera 4 via the image input unit 7 are taken into the memory 8 (# 01). The captured input image is adjusted for brightness (density value) by the brightness adjusting unit 61 (# 02). At that time, a feature amount of the input image is required, and it is preferable that the input image is divided into a predetermined number of sections and the maximum value of the density average value calculated for each section is used as the feature amount. The determination of the next binarization threshold of this feature can also be used to adjust the lens aperture of the imaging camera 4. The binarization threshold is determined by the binarization threshold determination unit 62a (# 03), and after the image smoothing and edge enhancement are performed by the image feature extraction unit 62b (# 04), this input image is binary. It is converted into a binarized image (# 05).

From the binarized input image, an isolated bright pixel region consisting of a predetermined number of pixels (predetermined from the image resolution, etc.) is extracted as a defect candidate by the depression candidate extraction unit 63 (# 06). .. Of the extracted defect candidates, defect candidates belonging to isolated points that are instantaneously and locally generated by ambient light or the like are excluded from the defect candidates by the defect candidate time series determination unit 64a (# 07), and further extracted defect candidates. Of these, defect candidates belonging to isolated points located in the interrupted region of the luminescent image are excluded from the defect candidates by the luminescent image discontinuous part search unit 64b (# 08).

The peripheral area including the interrupted area of the light emission image found by the light emission image discontinuous part search unit 64b is the shape information of the automobile body 1 as an object to be inspected transmitted from the host computer 14 and the transport position information by the conveyor 2. It is masked by the image mask generation unit 65 as an unnecessary pixel area together with a background area other than the painted surface as the surface to be inspected determined based on (# 09). In this embodiment, the transport position information obtained from the host computer 14 may differ from the actual position. Therefore, a laser sensor or the like is used to check the position shift of the automobile body 1 in real time, and the image thereof. The position of the mask is corrected (# 10).

After selecting the defect candidates and removing the background image in this way, the remaining defect candidates (isolated points) are labeled (# 11), and the area of the isolated points assigned to each label is calculated (# 11). # 12), only isolated points that satisfy the preset area condition (whether or not the area is equal to or larger than the threshold) are judged as true defects (# 13), and their coordinate positions and sizes are used in the defect map. Write (# 14).

This completes the procedure for evaluating defects on the painted surface by the defect evaluation means 6, but when the inspection of the painted surface is completed through this procedure, the controller 5 of the surface defect inspection device is used at the painted surface inspection verification station via the host computer 14. Of the defect maps sent from, defect matching is performed using a defect map that is given an ID that matches the ID of the automobile body carried into the painted surface inspection verification station. At that time, in order to facilitate the collation work by the inspector, it is convenient to operate the projector 15 so as to point out the defect portion based on the corresponding defect map. Of course, the defect information based on such a defect map may be output on paper by the printer 13 connected to the output unit of the surface defect inspection device, and this output paper may be directly attached to the automobile body 1.

In the above-described embodiment, the illumination unit 3 is composed of a group of LED elements continuously arranged in a hexagonal network. However, the network form may be other than a hexagon, and the light emitting element 30 may be an LED element. Other than may be adopted. Further, instead of arranging the image pickup camera 4 in the dark surface surrounded by the LED element group, the image pickup camera 4 may be arranged outside the illumination unit 3, but in that case, the shooting angle of the image pickup camera 4 with respect to the surface to be inspected. It is convenient to convert the input image into an image that accurately faces the irradiation surface 3a of the illumination unit 3 by correcting the input image by tilting based on the above.

In the above-described embodiment, the input image is converted into a binarized image for defect detection, but the present invention is not limited to the binarization of the input image, and the input image is binarized or more. Defect detection using a multi-valued image is also within the framework of the present invention.

<figref num="1">Schematic diagram of the surface defect inspection apparatus according to the present invention.</figref><figref num="2">Schematic diagram showing the lighting unit and the imaging camera</figref><figref num="3">Functional block diagram showing the configuration of defect evaluation means mounted on the surface defect inspection device</figref><figref num="4">Explanatory drawing explaining the binarized input image</figref><figref num="5">Explanatory drawing explaining the isolated point existing in the interrupted part of a light emitting image</figref><figref num="6">Flow chart showing the procedure of defect evaluation of the surface to be inspected by the defect evaluation means</figref>

Code description

3: Lighting unit 4: Imaging camera 5: Controller 6: Defect evaluation means 30: Light emitting element (LED element) 31: Dark surface 60A: Preprocessing unit 60B: Defect determination unit 61: Luminance adjustment unit 62: Binarization processing unit 63: Defect candidate (isolated point) extraction unit 64: Defect candidate selection unit 65: Image mask generation unit 66: Label setting unit 67: Area calculation unit 68: Defect determination unit