JP5481012B2 - Surface inspection device - Google Patents

Description

  The present invention relates to a surface inspection apparatus that performs surface inspections such as surface defects and color defects of an inspection object such as an automobile body using a camera having an image sensor.

  Conventionally, surface inspection of a vehicle body at an automobile manufacturing site has been performed visually by a large number of inspectors in order to surely detect even surface defects (unevenness) and slight defects in color.

On the other hand, various techniques for automating the surface inspection of the vehicle body have been proposed. For example, Japanese Laid-Open Patent Publication No. 2000-241147 (Patent Document 1) discloses a vehicle body transport direction for in-line inspection of surface defects such as irregularities and protrusions with a gentle inclination angle of a vehicle body surface caused by press forming of the vehicle body. A wide-angle illuminating unit that irradiates the surface to be inspected with a certain irradiation angle α from before and after the light, and a reflected light of the illuminating light at an imaging angle β larger than α before and after the wide-angle illuminating unit A technique is described in which imaging means for imaging are arranged and surface defects on the surface to be inspected are extracted based on received light images obtained by the preceding and following imaging means.
JP 2000-241147 A

  Visual inspection of the surface requires a lot of labor and time. In addition, in order to detect defects visually, it is necessary to perform inspection while passing the vehicle body through a so-called light tunnel in which a large number of illumination light sources are arranged on both sides, which increases the equipment cost.

  With the technique described in Patent Document 1, it is possible to inspect a shape defect such as unevenness, but it is not possible to inspect a color defect such as color unevenness.

  An object of this invention is to provide the surface inspection apparatus which can test | inspect simultaneously the surface state regarding the shape and color of a test subject like a vehicle body.

A surface inspection apparatus according to an aspect of the present invention includes a line sensor that scans the surface of an inspection object in a main scanning direction at a constant cycle, a camera that outputs image data, and a secondary sensor that is orthogonal to the main scanning direction. The sub-scanning means for relatively moving the camera and the inspection object in the scanning direction, and the main scanning area of the surface of the inspection object, the illumination angles differing from the normal line rising from the surface Illuminating means including a plurality of light sources arranged so as to illuminate, driving means for selectively lighting at least one of the plurality of light sources by switching, and for each selective lighting of the light sources by the driving means Corresponding to each selective lighting of each of the light sources, arithmetic processing means for performing processing on the image data output from the camera and outputting a processing result indicating the surface state of the inspection object Display means for individually displaying the processing results, wherein the plurality of light sources have a relatively large illumination angle from the normal line and a relatively large illumination angle from the normal line. A plurality of first light sources and a plurality of second light sources, each of which is arranged to irradiate the same irradiation position of the inspection object from different directions; Selectively turns on each of the plurality of light sources at a lighting cycle that is an integral multiple of a main scanning cycle synchronized with scanning in the main scanning direction by the line sensor, and the arithmetic processing means is the first driving means by the driving means. When the light source is turned on, a processing result indicating the surface state related to the shape is output. When the second light source is turned on, a processing result indicating the surface state related to the color is output. Simultaneous to display an image corresponding to each light source obtained when the different display areas of one screen.

  According to the present invention, the main scanning region on the surface of the object to be inspected is sequentially illuminated at different illumination angles, and the processing results at each illumination angle are individually generated and displayed, so depending on a plurality of different illumination angles The surface state relating to the shape and color of the inspection object, that is, the shape and color defects on the surface of the inspection object can be simultaneously inspected.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 shows a surface inspection apparatus according to an embodiment of the present invention. The inspection object 1 is not limited to this, but is, for example, a car body. The inspection object 1 is placed on the conveyance path 2 at the time of inspection, and is moved together with the conveyance path 2 in the direction indicated by the arrow Y (sub-scanning direction) by a sub scanning device (not shown). A camera (digital camera) 3 for photographing the side surface of the inspection object 1 is installed on the side of the conveyance path 2. In addition to the camera 3, you may install the camera which image | photographs the upper surface of the test target object 1 further.

  The illumination device 4 is arranged in front of the camera 3, that is, on the side close to the inspection object 1 with respect to the camera 3. The illumination device 4 includes a plurality of linear light sources 5A, 5B, 5C, and 5D and a support member 6 that supports the light sources 5A, 5B, 5C, and 5D as shown in detail in FIG. Each of the light sources 5A, 5B, 5C, and 5D is, for example, an LED array in which a plurality of LEDs (light emitting diodes) are arranged in a line. The illumination device 4 is installed so that the array direction of the LED array is substantially vertical. Note that a discharge tube such as a cold cathode discharge tube may be used instead of the LED array.

  The support member 6 has, for example, a semi-cylindrical shape, and the light sources 5A, 5B, 5C, and 5D are supported on the inner surface thereof. In a region between the light sources 5 </ b> B and 5 </ b> C of the support member 6, a slit S is provided for allowing the reflected light from the surface of the inspection object 1 to pass and guide it to the camera 3. The support member is not limited to such a shape, and may be one that supports both sides of the light sources 5A, 5B, 5C, and 5D, for example. In this case, the slit S is not necessary, and it is only necessary to guide the reflected light from the surface of the inspection object 1 to the camera 3 through the gap between the light sources 5A, 5B, 5C and 5D.

  Here, the light sources 5 </ b> A, 5 </ b> B, 5 </ b> C, and 5 </ b> D illuminate a linear region (referred to as a main scanning region of the camera 3) L on the surface of the inspection object 1 from different angles with respect to the main scanning region L. Are arranged as follows. That is, among the light sources 5A, 5B, 5C, and 5D, the illumination angles of the two light sources 5A and 5D on both sides are relatively large, and the illumination angles of the other light sources 5B and 5C are relatively small. Here, the illumination angle is an angle of the illumination direction (the optical axis direction of illumination light) with respect to the front direction (0 ° direction) of the main scanning region L. As described above, the illumination device 4 illuminates the main scanning region L from the oblique direction with a large illumination angle and the main scanning region L from a direction near the front with a small illumination angle (an angle close to vertical). Light sources 5B and 5C are provided. Here, the illumination device 4 includes four light sources 5A, 5B, 5C, and 5D. However, a larger number of light sources may be arranged with different illumination angles.

  The camera 3 includes an imaging lens 11, a line image sensor 12 such as a CCD line image sensor, an amplifier 13, and an A / D converter 14. An image of the surface of the inspection object 1 is lined through the imaging lens 11. An image is formed on the image sensor 12. The line image sensor 12 is configured by arranging a plurality (for example, 5120) of photoelectric conversion elements in a direction (main scanning direction) indicated by an arrow X orthogonal to the sub-scanning direction Y. The line image sensor 12 scans the surface of the inspection object 1 in the main scanning direction X with a period of 2,000 scans / second (main scanning period), for example, under the control of the controller 9. The state is read as an image and an image signal is output. When the inspection object 1 is an automobile body, the distance (objective distance) from the camera 3 to the inspection object 1 is set to about 3,600 mm, for example. Further, the length of the main scanning region L (referred to as main scanning width) is set to about 2,400 mm, for example.

  The image signal output from the line image sensor 12 is amplified by the amplifier 13, further converted to, for example, 8-bit parallel digital data by the A / D converter 14, and output from the camera 3 as image data.

  A light source drive circuit 8 is connected to the light sources 5A, 5B, 5C, and 5D of the illumination device 4 via a light source switch 7. The light source switch 7 is controlled by the controller 9, and sequentially turns on the light sources 5A, 5B, 5C, and 5D at a lighting cycle that is an integral multiple of the main scanning cycle in synchronization with the main scanning of the line image sensor 12. That is, the light source switch 7 is switched in a lighting cycle that is an integral multiple (for example, 1 time) of the main scanning period in synchronization with the main scanning, and the driving current output from the light source driving circuit 8 is selectively changed to the light sources 5A, 5B, Supply to 5C and 5D. Since the LED array and the cold cathode discharge tube have a very fast response speed, it is easy to blink the light sources 5A, 5B, 5C and 5D at a fast cycle as described above.

  The main scanning region L is illuminated by one of the light sources 5A, 5B, 5C, and 5D that is lit. The camera 3 captures the main scanning area L by detecting the reflected light from the illuminated main scanning area L. Image data output from the camera 3 is input to the arithmetic processing unit 16. The arithmetic processing device 16 outputs a processing result indicating the surface state of the inspection object 1 by performing arithmetic processing as will be described later on the input image data. The processing result output from the arithmetic processing unit 16 is supplied to a display device 16 such as a liquid crystal display or a CRT display and displayed as an image.

The arithmetic processing unit 16 performs arithmetic processing in synchronization with the main scanning by the line image sensor 12 under the control of the controller 9, in other words, for each lighting of the light sources 5A, 5B, 5C and 5D by the light source switch 7. Do. The processing results obtained for each lighting of the light sources 5A, 5B, 5C and 5D are individually displayed in the display device 16, that is, in different areas 10A, 10B, 10C and 10D on the screen of the display device 16 as shown in FIG. Displayed as an image.

  In general, when observing the smooth surface of an object, (a) the surface shape can be recognized by observing from an oblique direction, but depending on the angle, the color of the light source is observed due to reflection of the light from the light source. The actual color of the surface cannot be recognized. On the other hand, (b) when the surface of the object is observed from the front or near the direction of the angle, the actual color can be recognized, but it is difficult to recognize the shape such as the unevenness.

According to one embodiment of the present invention, when the light sources 5A and 5D having a large illumination angle are turned on, a processing result indicating the surface state relating to the shape of the inspection object 1 is output from the arithmetic processing device 16 according to the principle (a). The When the light sources 5B and 5C having a small illumination angle are turned on, a processing result indicating the surface state relating to the color is output from the arithmetic processing unit 16 from the principle of (b). Therefore, by displaying these processing results individually as images as shown in FIG. 4, it is possible to simultaneously observe the surface state regarding the shape and color.

  In the above description, the light sources 5A, 5B, 5C, and 5D are turned on one by one. However, a mode in which a plurality of light sources are turned on simultaneously may be included. For example, (1) Only 5A is lit, (2) 5A and 5B are lit simultaneously, (3) Only 5B is lit, (4) 5B and 5C are lit simultaneously, (5) Only 5C is lit, (6) 5C And 5D may be turned on at the same time. In this way, four processing results are output from the arithmetic processing unit 16 for one main scanning region L (when the light sources 5A, 5B, 5C and 5D are turned on one by one, six processing results are output. Main scanning is performed from a larger number of directions.

  The car body of the automobile which is the inspection object 1 has various colors. Even in the same car body, the color differs between the so-called white body before painting and after painting. If the colors are different, the lightness will be different, and the intensity of the reflected light when the main scanning region L is illuminated with the light sources 5A, 5B, 5C and 5D will also be different. In addition, the body of an automobile has various surface roughnesses. For example, the surface of the automobile is glossy and has a mat shape. When the surface is glossy, the intensity of the reflected light incident on the camera 3 is high. When the surface is matte, the intensity of the reflected light incident on the camera 3 decreases because the light diffuses.

  Therefore, the controller 9 changes the driving current output from the light source driving circuit 8 in accordance with the surface color (brightness) and surface roughness (glossy or matte) of the inspection object 1, and the surface brightness is changed. If it is high or the surface is glossy, the drive current is lowered to reduce the light emission intensity of the light sources 5A, 5B, 5C and 5D. If the lightness is low or the surface is rough like a mat, the light emission intensity may be increased. desirable. Thus, by changing the light emission intensity of the light sources 5A, 5B, 5C and 5D according to the color (brightness) of the inspection object 1 and the surface roughness, the inspection can be performed with a substantially constant sensitivity regardless of the brightness and roughness. It can be carried out.

  Further, when the inspection object 1 is a car body of an automobile, the surface shape is not flat, generally a curved surface, and also changes in the length direction. That is, the shape of the main scanning region L is not constant and varies depending on the inspection object 1 and the position in the length direction of the inspection object 1. Therefore, it is desirable to make the shapes of the light sources 5A, 5B, 5C and 5D variable. Specifically, for example, the light sources 5A, 5B, 5C, and 5D of the illumination device 4 and the support member 6 are made flexible and matched to the shape of the main scanning region L according to the position of the inspection object 1 or the main scanning region L. Thus, the shapes of the light sources 5A, 5B, 5C and 5D and the support member 6 may be changed.

  Next, the arithmetic processing unit 16 will be specifically described. In the camera 3 using the line image sensor 12, the lightness reading accuracy by the image sensor 12 varies depending on the photoelectric conversion element. This variation is caused by a difference in sensitivity among the individual photoelectric conversion elements constituting the image sensor 12, which is called element variation and normally takes a value of about 3%. In a general surface inspection apparatus, a fine defect or the like cannot be detected unless there is a change in brightness exceeding the range of element variation. Since it is said that the visual light / dark detection accuracy is 1/1500 to 1/2000, the surface inspection device using the line image sensor has only 1/60 of the visual accuracy, and cannot be substituted for the visual inspection. It has been said.

  As an example of a method for compensating for the element variation of the image sensor to solve this problem, the image data of a block composed of a plurality of pixels continuous in the main scanning direction in the image data sequence obtained by the line image sensor is added, and A technique for performing a correlation operation on addition data of adjacent blocks is known. However, this method can detect a defect that extends over adjacent pixels on one main scanning line, but it is possible to detect a defect that extends across adjacent main scanning lines or adjacent on the main scanning line. It is not possible to detect a defect that straddles pixels and straddles adjacent main scanning lines. For this reason, there is a limit in improving inspection accuracy.

  The surface inspection apparatus including the arithmetic processing unit 16 described here has a defect on the inspection object that extends over adjacent main scanning lines, adjacent main scanning lines, and adjacent main scanning lines. It is possible to detect defects that straddle. Accordingly, as described above, the main scanning region L on the surface of the inspection object 1 is sequentially illuminated by the illumination device 4 at different illumination angles, and the processing result at each illumination angle is individually generated and displayed. In addition, it is possible to accurately inspect the surface condition regarding colors.

  Furthermore, the arithmetic processing unit 16 described here can have a resolution as high as, for example, a diameter of 100 μm for a point and a width of about 20 μm for a line (the width of the hair can be identified). This is about the same resolution as when an inspection is performed by approaching the object up to about 500 mm. Accordingly, with regard to the depth of field (the range in which the in-focus state can be obtained), the object distance from the camera 3 to the inspection object 1 is 3,600 mm and the main scanning width is 2,400 mm. The depth of field is as large as about 1,300 mm. For this reason, the camera 3 placed on a fixed position relatively far from the inspection object 1 such as an object distance of 3,600 mm is used for the surface state of the inspection object having a rough surface like an automobile body. It is fully possible to inspect.

  In this embodiment, the light sources 5A, 5B, 5C, and 5D are arranged on the side closer to the inspection object 1 from the camera 3, but the positional relationship between the camera 3 and the light sources 5A, 5B, 5C, and 5D is limited to this. It is not a thing. For example, the camera 3 is arranged on the side of the conveyance path 2 and the reflected light from the main scanning region L on the surface of the inspection object 1 is guided to the camera 3 by a mirror arranged in front of the light sources 5A, 5B, 5C and 5D. It may be.

Next, some specific examples of the arithmetic processing unit 16 will be described.
(First specific example of arithmetic processing unit)
The arithmetic processing unit 16 according to the first specific example includes a line memory 17, an adder 18, an arithmetic unit 19, and a determination unit 20 as shown in FIG. 4. The line memory 17 converts the image data input from the camera 3 into a plurality (n) of main data corresponding to the number of light sources 5A, 5B, 5C, and 5D (strictly speaking, the number of scans of one main scanning region L). A memory for storing scan lines, and a shift register or FIFO (first-in / first-out) memory is used.

  Image data output from the camera 3 is supplied to the first input terminal of the adder 18 and also supplied to the line memory 17. The output of the line memory 17 is supplied to the second input terminal of the adder 18. The adder 17 adds the input image data and the output image data of the line memory 17.

  Here, the image data respectively input to the first input A and the second input B of the adder 18 corresponds to an image signal obtained from the same element of the line image sensor 12. That is, the input B is delayed by the time corresponding to n main scanning lines by the line memory 17 with respect to the input A of the adder 18. For example, when image data corresponding to the i-th (i = 1, 2,...) Element of the line image sensor 12 is input to the input A, one main scanning line corresponding to the same i-th element is input to the input B. Previous image data is input from the line memory 17. Accordingly, in the adder 18, as shown in FIG. 5, two main scanning lines temporally adjacent to each of the inputs A and B (the Nth line and the N + 1th line, the N + 1th line and the N + 1th line,...) Are input. Are added to generate an image data string 22. Here, the two main scanning lines that are temporally adjacent are data that are temporally adjacent to the image data output from the camera 3 in response to lighting of the same light source. For example, the first image data output from the camera 3 corresponding to the lighting of the light source 5A and the second image data output from the camera 3 corresponding to the next lighting of the same light source 5A are temporally adjacent. Suppose that The same applies to the following specific examples.

  The image data sequence 22 thus generated by the image data sequence generator constituted by the line memory 17 and the adder 18 is input to the arithmetic unit 19. The computing unit 19 adds (accumulates) the image data of the block composed of a plurality of pixels continuous in the main scanning direction X to the image data string from the adder 18 to generate in-block addition data, The process of calculating the correlation value of the intra-block addition data adjacent to each other in the main scanning direction and the pixel data of the first pixel is repeatedly performed by shifting the block position in the main scanning direction.

  Specifically, as shown in FIG. 6, the arithmetic unit 19 includes an M-stage shift register 31 and a 2M-stage shift register 31 connected so that the image data sequence 22 (see the figure) from the adder 18 is input to the first stage. The shift register 32, an adder 33 for adding the outputs of the respective stages of the shift register 31, an adder 34 for adding the outputs of the M stages subsequent to the shift register 32, and the outputs of the adders 33 and 34 are supplied. The correlator 35 is used. Here, when the pixel of the image data of one main scanning line is divided into blocks composed of a plurality of pixels continuous in the main scanning direction, M is the number of pixels constituting one block. It is desirable that the value of M can be arbitrarily changed. For example, the value of 1-111 is taken.

  The operation of the arithmetic unit 19 will be described with reference to FIG. The adder 33 adds the image data of one block composed of M pixels continuous in the main scanning direction in the image data string 22 output from the adder 18. The adder 34 adds the image data of the next block adjacent in the main scanning direction to the block to which the image data is added by the adder 33 in the image data sequence 22. Here, if the intra-block addition data output from the adders 33 and 34 are b1 and b2, the correlator 35 obtains, for example, the difference b1−b2 between them as the correlation value 36.

  Each time new pixel data in the image data string 22 is input to the shift registers 31 and 32, the positions of the blocks to which the image data is added by the adders 33 and 34 are sequentially shifted in the main scanning direction as shown in FIG. Shifted and the same operation is performed. By such operations, the adders 33 and 34 sequentially output the intra-block addition data c1, c2; d1, d2; e1, e2;..., And the correlator 35 outputs c1-c2, d1-d2, e1-e2. Are sequentially obtained as the correlation value 36.

  Here, the correlator 25 calculates the difference between the adjacent intra-block addition data as the correlation value 36, but the ratio (b 1 / b 2,...) Of the adjacent intra-block addition data may be calculated as the correlation value 36. The correlation value 36 output from the correlator 25 is input to the determiner 20 in FIG. The determiner 20 is configured by a comparator, for example, and compares the correlation value 36 output from the correlator 26 with an appropriate threshold value to determine the presence / absence of a surface defect on the inspection object 10 and to indicate a surface state. Output the result.

  That is, if there is a defect on the inspection object 10, the size of the image data corresponding to the same element of the line image sensor 12 in the vicinity of the defect is accompanied by sub-scanning, that is, the relative movement of the inspection object 10. Accordingly, the correlation value 36 obtained by the correlator 35 increases and exceeds the threshold value in the determiner 20, so that the defect can be recognized by the determiner 20. The determination result (processing result) of the determination device 20 is processed by, for example, a personal computer and displayed on a display device (not shown).

  According to the arithmetic processing unit 15 configured as described above, first, an accumulation function by performing intra-block addition in the arithmetic unit 19 and an autocorrelation function by correlating adjacent intra-block addition data in the main scanning direction. In addition to removing the influence of element variations of the line image sensor 12, the correlation value 36 obtained by the correlator 35 is increased, and the defect detection sensitivity can be increased.

  Further, the image data of two main scanning lines adjacent in the sub-scanning direction are added using the line memory 17 and the adder 18 in particular, and an image data string obtained thereby is input to the arithmetic unit 19. Of course, when the defect 51 on the inspection object 10 exists only in one main scanning line (Nth line in the figure) as shown in FIG. 8A, two adjacent main scanning lines (FIG. 8B) In FIG. 5C, a defect 52 that straddles adjacent pixels on the main scanning line and straddles two adjacent main scanning lines as shown in FIG. 5C. As a result of the detection, it is possible to perform inspection with higher accuracy.

Hereinafter, this principle will be described in detail.
First, when the defect 51 on the inspection object 10 exists only in one main scanning line as shown in FIG. 8A, image data (FIG. 9A) corresponding to the defect 50 existing in one pixel ( Compared with (camera 3 output), the output of the image data corresponding to the defect 51 existing across two pixels becomes smaller. Since the data for one pixel is distributed to two pixels, the output level is halved. Since there is no defect in the next line, the output is 0, and the output of the adder 18 (image data string 22) for adding the image data of two lines adjacent in the sub-scanning direction is the same as the image signal of the N line. . The pixel data in the block in the main scanning direction is accumulated in the image data string 22 by the accumulation function of the computing unit 19 to obtain predetermined, for example, first pixel data in the block.

  Here, for convenience of explanation, assuming that the block consists of two pixels, the adder 33 adds two image data strings shifted by one pixel as shown in FIG. 9B. Therefore, the next pixel data is added to each pixel data, and the level of the image data of the pixel c6 ′ corresponding to the defect 51 is the same as the image data of the pixel c2 ′ corresponding to the defect 50 (level). Therefore, not only the defect 50 but also the defect 51 can be easily detected.

  Next, as shown in FIG. 8B, when there is a defect 52 extending over two adjacent main scanning lines, it corresponds to the defect 50 existing in one pixel in one main scanning line as shown in FIG. 10A. The output of the image data corresponding to the defect 52 is smaller than the image data to be processed. This is the same for both the N line and the (N + 1) line. However, the output of the image data sequence 22 corresponding to the defect 52 becomes the same as the output of the image data sequence 22 corresponding to the defect 51 by the adder 18 that adds the image data of two lines adjacent in the sub-scanning direction. For this reason, it is possible to easily detect the defect 52 existing across the two main scanning lines. FIG. 10B shows intra-block addition data, but in this example, no intra-block addition is necessary because there is no defect across two pixels in the main scanning direction.

  Furthermore, when there is a defect 53 that spans adjacent pixels on the main scan line and extends across two adjacent main scan lines as shown in FIG. 8C, one defect in one main scan line is shown in FIG. 11A. The output of the image data corresponding to the defect 53 is smaller than the image data corresponding to the defect 50 existing in one pixel. Since the data for one pixel is distributed to four pixels, the output level of the image data corresponding to the defect 53 is 1/4 for both the N line and the (N + 1) line. However, the output level of the image data string 22 corresponding to the defect 53 is amplified to ½ by the adder 18 that adds two lines of image data adjacent in the sub-scanning direction. The image data string 22 is converted into pixel data of a predetermined, for example, first pixel in the block by the effect of the accumulation function in the main scanning direction of the calculator 19.

  Here, for convenience of explanation, assuming that the block consists of two pixels, the adder 33 adds two image data strings shifted by one pixel as shown in FIG. 11B. Therefore, the next pixel data is added to each pixel data, and the level of the image data of the pixel c6 ′ corresponding to the defect 53 is the same as the image data of the pixel c2 ′ corresponding to the defect 50 (level). Therefore, not only the defect 50 but also the defect 53 can be easily detected.

(Second specific example of the arithmetic processing unit)
Next, a second specific example of the arithmetic processing unit 16 will be described with reference to FIG. In the first specific example described above, the image data 21 output from the camera 3 is added to the image data of two main scanning lines that are temporally adjacent using the line memory 17 and the adder 18. When the data string 22 is generated, image data of two pixels adjacent to the two main scanning lines in the sub-scanning direction Y (pixels having the same position in the main scanning direction) are added.

  However, the method of generating the image data string by adding the image data of two main scanning lines adjacent in the sub-scanning direction is not limited to this. For example, the image data string is generated by the method shown in FIG. It may be generated. In this method, for the image data 21 of two main scanning lines adjacent in the sub-scanning direction Y, the process of adding the image data of four pixels adjacent in the sub-scanning direction and adjacent in the main scanning direction is the four pixels. Is shifted in the main scanning direction in units of one pixel to generate the image data string 22A.

  According to this method, since the size of each data in the image data sequence is increased, the inspection accuracy can be further increased, and the image data straddles adjacent pixels on the main scanning line as described with reference to FIGS. 8C and 11A, for example. In addition, it is particularly effective in detecting a defect 53 existing across adjacent main scanning lines.

(Third specific example of the arithmetic processing unit)
In the third specific example, when the image data strings are generated by adding the image data of two main scan lines that are temporally adjacent, as shown in FIG. Image data of two pixels (pixels having the same position in the main scanning direction) adjacent to each other in the sub-scanning direction of the two main scanning lines (here, the N line and the N + 1 line are used for convenience) In addition to the processing to be generated, the processing in FIGS. 13B and 13C is used in combination. FIG. 13B is a process for generating an image data string by adding image data of adjacent pixels in a direction (first direction) obliquely upward to the sub-scanning direction. FIG. 13C is a process for generating an image data string by adding image data of adjacent pixels in a direction obliquely upward to the left (second direction) with respect to the sub-scanning direction.

  FIG. 14 is a diagram showing the configuration of the arithmetic processing unit 16A in the third specific example. The line memory 17A is composed of the number of pixels for one main scanning line (in this example, 5120 pixels) + 1 = 5121 stages of shift registers. Composed. The input of this shift register, the 5120th stage output, the 5121st stage output, and the 5119th stage output are added by the adders 18A, 18B, and 18C, respectively. As a result, the adder 18A adds the image data of two pixels adjacent to each other in the sub-scanning direction of the two main scanning lines shown in FIG. 10A, and the adder 18B in the sub-scanning direction shown in FIG. 13B. An image data string obtained by adding the image data of pixels adjacent in the diagonally upward direction, and the image data of the pixels adjacent in the diagonally upward direction as shown in FIG. 13C from the adder 18C. An image data string obtained by adding is output.

  These image data strings are respectively input to the determiners 20A, 20B, and 20C through the same arithmetic units 19A, 19B, and 19C as described in the first specific example, and are subjected to threshold determination. The determination results of these determiners 20A, 20B, and 20C are processed by, for example, a personal computer and displayed on a display device (not shown). In this case, the determination results of the determiners 20A, 20B, and 20C may be displayed in different colors so that they can be distinguished from each other.

  According to the third specific example, for example, a very thin defect such as a line existing on the inspection object 10 and a defect that crosses the line image sensor 12 at an angle is easily detected. Is possible. That is, such a defect appears across two adjacent pixels in the diagonal direction with respect to the sub-scanning direction as shown in FIG. 13B or 13C on the two main scanning lines, and the image data of these pixels are added. After that, it is easily detected by processing with an arithmetic unit.

  In the description so far, the image data sequence obtained by adding the image data of two main scanning lines temporally adjacent to each other using the line memory and the adder is subjected to intra-block addition via the arithmetic unit, and then to the determination unit. Although it is input, it may be input to the determiner without going through the calculator.

  The human eyeball performs a minute tremor called fixation fixation. That is, it is considered that the eyeball improves accuracy by performing fixation fixation fine movement separately from the movement when looking up, down, left, and right, thereby preventing the retina from getting used to stimulation. The main movement component of fixation fine movement is the vertical direction.

  The process of adding the image data of two adjacent main scanning lines corresponds to such a human eyeball fine movement. That is, in the present embodiment, surface inspection is performed by processing an image data sequence obtained by adding image data of two adjacent main scanning lines while performing movement (fine movement) in the sub-scanning direction by electronic processing or mechanical processing. By doing this, the inspection accuracy is increased.

  According to the arithmetic processing apparatus as described above, a defect that extends over the main scanning line on the inspection object 1 or a defect that extends over adjacent pixels on the main scanning line and extends over adjacent main scanning lines. It is possible to detect. Therefore, coupled with the fact that the main scanning region L on the surface of the inspection object 1 is sequentially illuminated by the illumination device 4 at different illumination angles and the processing results at the respective illumination angles are individually generated and displayed. Inspection of the surface state regarding the shape and color can be performed with higher accuracy.

The block diagram which shows the structure of the surface inspection apparatus which concerns on one Embodiment of this invention. The perspective view which shows the detail of the illuminating device in FIG. The figure which shows the example of an image display in the display apparatus in FIG. The block diagram which shows the 1st specific example of the arithmetic processing unit in FIG. FIG. 6 is a diagram for explaining an operation of generating an image data string by adding image data of two main scanning lines adjacent in the sub-scanning direction in the embodiment. The block diagram which shows the structure of the calculator in FIG. The figure explaining operation | movement of the calculator shown in FIG. The figure explaining the example of the defect on the inspection object A diagram explaining examples of other defects on the inspection object The figure explaining the example of another defect on a test subject The figure which shows the relationship between the defect which exists only in one main scanning line, the image data corresponding to it, and the addition data in a block The figure which shows the relationship between the image data corresponding to the defect which exists only in one main scanning line, and the addition data in a block The figure which shows the relationship between the defect which exists over two adjacent main scanning lines, the image data corresponding to it, and the addition data in a block The figure which shows the relationship between the image data corresponding to the defect which exists over two adjacent main scanning lines, and the addition data in a block FIG. 4 is a diagram showing the relationship between a defect that spans adjacent pixels on the main scanning line and spans two adjacent main scanning lines, and corresponding image data and intra-block addition data. FIG. 4 is a diagram showing the relationship between image data and intra-block addition data corresponding to a defect that spans adjacent pixels on the main scanning line and spans two adjacent main scanning lines. The figure explaining the operation | movement which adds the image data of two main scanning lines adjacent in the subscanning direction in the 2nd specific example of an arithmetic processing unit, and produces | generates an image data sequence. The figure explaining the modification which adds the image data of two main scanning lines adjacent in the subscanning direction in the 3rd specific example of an arithmetic processing unit, and produces | generates an image data sequence. The figure explaining the modification which adds the image data of two main scanning lines adjacent in the subscanning direction in the 3rd specific example of an arithmetic processing unit, and produces | generates an image data sequence. The figure explaining the modification which adds the image data of two main scanning lines adjacent in the subscanning direction in the 3rd specific example of an arithmetic processing unit, and produces | generates an image data sequence. Block diagram showing a third specific example of the arithmetic processing unit

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Inspection object 2 ... Conveyance path 3 ... Camera 4 ... Illumination device 5A, 5B, 5C, 5D ... Light source 6 ... Support body 7 ... Light source switcher 8. ··· Light source drive circuit 9 ··· Controller 10A, 10B, 10C, 10D · · · Display area 11 · · · Imaging lens 12 · · · Line image sensor 13 · · · Amplifier 14 · · · A / D converter 15 ... Operation processing unit 16 ... Display device

Claims (2)

A camera that has a line sensor that scans the surface of the inspection object in the main scanning direction at a constant cycle, and that outputs image data;
Sub-scanning means for relatively moving the camera and the inspection object in a sub-scanning direction orthogonal to the main scanning direction;
Illuminating means including a plurality of light sources arranged so as to illuminate the main scanning region of the surface of the object to be inspected at illumination angles different from each other from the normal rising from the surface;
Driving means for selectively lighting at least one by switching the plurality of light sources;
Arithmetic processing means for performing arithmetic processing on image data output from the camera for each selective lighting of the light source by the driving means and outputting a processing result indicating a surface state of the inspection object;
Display means for individually displaying the processing result corresponding to each selective lighting of the light source,
The plurality of light sources include a first light source having a relatively large illumination angle from the normal and a second light source having a relatively small illumination angle from the normal,
And there are a plurality of the first light source and the second light source, each of which is arranged to irradiate the same irradiation position of the inspection object from different directions,
The driving means selectively turns on each of the plurality of light sources at a lighting cycle that is an integral multiple of a main scanning cycle synchronized with scanning in the main scanning direction by the line sensor,
The arithmetic processing means outputs a processing result indicating a surface state relating to a shape when the first light source is turned on by the driving means, and outputs a processing result indicating a surface state relating to a color when the second light source is turned on.
The surface inspection apparatus, wherein the display means simultaneously displays an image corresponding to each light source obtained when each of the plurality of light sources is irradiated on different display areas of one screen.   Each of the first and second light sources is an LED array in which a plurality of LEDs are arranged in a line, and is supported on the inner surface of a semi-cylindrical support member and in the line in the axial direction of the cylinder. The surface inspection apparatus according to claim 1.

Images