JP2010197177A - Visual inspection method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce time and labor by automatically setting a division area required for the verification of the continuity of cracks when the cracks have occurred in an object to be inspected. <P>SOLUTION: An edge image is generated at an edge image generating part 12 on the basis of an original image, a grayscale image acquired by imaging the object to be inspected 1 by a TV camera 2. A processing part 15 detects a crack candidate from the edge image. A division area generating part 17 divides an inspection region into a plurality of small sections by auxiliary division lines passing through an inspection start point set on a boundary line of the inspection region and generates division areas by further dividing each small section in different directions. A crack determination part 18 determines the crack candidate as cracks when the number of division areas over which the crack candidate extends is equal to or greater than a threshold. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an appearance inspection method and an appearance inspection apparatus for detecting a crack generated in an inspection object by an image processing technique.

  2. Description of the Related Art Conventionally, a technique for inspecting the appearance of an object to be inspected in an image by performing image processing by a computer on an image acquired using an imaging means such as a TV camera is known. For example, after imaging a spatial region including the object to be inspected by the image pickup means, the contour line of the object to be inspected using the change in the lightness level (density) of each pixel of the original image that is the light and shade image obtained by the image pickup means A technology that inspects the inside of the outline of the inspection object from the original image and the edge image, and inspects the inspection object for chipping, cracks, and foreign matter (dust). Yes (see Patent Document 1).

  In the technique described in Patent Document 1, an inspection start point is set on the boundary line of the inspection region in the edge image, and a detection area that is linear and has a constant width is set through the inspection start point. Rotating around the inspection start point. That is, the detection area is rotated and scanned in the inspection area.

  Whether or not the inspection object is cracked is determined by using the differential absolute value and the differential direction value of the pixel in the detection area at each position when the detection area is rotationally scanned. That is, when a pixel whose differential absolute value exceeds a specified value and whose differential direction value is within a predetermined angle range with respect to the angle of the detection area is present in the detection area, that pixel is used as a defect candidate point. When the total number of defect candidate points and the sum of the differential absolute values of the defect candidate points exceed a prescribed threshold, it is determined that a crack exists in the inspection object.

  Patent Document 1 also describes a technique for verifying the continuity of cracks in order to make a more reliable determination on the cracks detected by the above technique. That is, the inspection area is divided into a plurality of divided areas, and it is verified whether or not a pixel group determined to be a crack exists across the plurality of divided areas.

JP 2007-147407 A

  By the way, in Patent Document 1, a rectangular divided area is nested in an inspection area, the inspection area is divided in the horizontal direction to detect a laterally extended crack, and the vertical direction is extended. Dividing the inspection area in the vertical direction in order to detect cracks generated, and dividing the inspection area in both the vertical and horizontal directions in order to detect both cracks extended in the vertical direction and cracks extended in the horizontal direction. Are listed.

  However, in the technique described in Patent Document 1, the division method of the divided area is empirically determined according to the crack generation direction, and it is considered that an appropriate divided area is automatically set regardless of the crack. Absent.

  The present invention has been made in view of the above-mentioned reasons, and its purpose is to reliably detect crack continuity and verify the crack continuity when a crack is generated in an inspection object. It is an object of the present invention to provide an appearance inspection method and an appearance inspection apparatus that save labor by automatically setting the necessary divided areas.

  The first aspect of the invention is an appearance inspection method, wherein a first step of setting an inspection start point on a boundary line of an inspection region set for an edge image generated from an original image that is a grayscale image of an inspection object; First, a detection area set linearly and at a constant width passing through the inspection start point is rotated around the inspection start point in the inspection area, and a pixel group that is a crack candidate is extracted at each position of the detection area. Set the auxiliary dividing line to divide the inspection area into a plurality of subdivisions through two steps and the inspection start point, and divide each subdivision in different directions by dividing lines passing through the plurality of dividing points on the auxiliary dividing line The third step of generating the divided areas, and there are crack candidates that extend beyond the specified number of the divided areas. And having a fourth step of determining the crack candidate cracks Rutoki.

  According to a second aspect of the present invention, in the first aspect of the present invention, the dividing line is a part of a polygonal outline centered on the inspection start point.

  In the invention of claim 3, in the invention of claim 1, the inspection area is rectangular and the inspection start point is set at one vertex of the inspection area, and in the third step, the inspection start point passes through the inspection start point. The inspection area is divided into two small sections by the auxiliary dividing line intersecting at 45 degrees with respect to the vertical axis of the edge image, and the inspection area is sandwiched between the plurality of dividing points set in the auxiliary dividing line. The inspection region is divided into a plurality of divided areas by perpendicular lines respectively extending on the two sides.

  According to a fourth aspect of the present invention, in the first aspect of the invention, the inspection area is rectangular and the inspection start point is set at one vertex of the inspection area. In the third step, the inspection passes through the inspection start point. The inspection area is divided into two subdivisions by the auxiliary dividing line with the diagonal of the area as the auxiliary dividing line, and each of the two sides of the inspection area sandwiching the inspection start point from the plurality of dividing points set in the auxiliary dividing line. The inspection area is divided into a plurality of divided areas by the lowered vertical line.

  In the invention of claim 5, in the invention of claim 1, the inspection area is rectangular and the inspection start point is set on one side of the inspection area. In the third step, the inspection area passes through the inspection start point. Two inspection lines are divided into three subdivisions by two auxiliary dividing lines with the two straight lines connecting the opposite end points of the two sides as the auxiliary dividing lines, and a plurality of divisions set for each auxiliary dividing line The inspection area is divided into a plurality of divided areas by a perpendicular line extending from the point to the remaining two sides of the inspection area and a straight line connecting the dividing points on each auxiliary dividing line.

  According to a sixth aspect of the present invention, in any one of the third to fifth aspects of the present invention, an intersection of a plurality of concentric circles centered on the inspection start point and the auxiliary dividing line is used as the dividing point. .

  The invention according to claim 7 is the invention according to any one of claims 3 to 5, wherein the number of divisions is set for each of the small sections in the divided area.

  The invention of claim 8 is characterized in that, in the invention of any one of claims 1 to 7, in the second step, an angle range for rotating the detection area is a range specified in the inspection area. And

  The invention of claim 9 is the invention according to any one of claims 1 to 8, wherein in the first step, a plurality of the inspection start points are set in the inspection area, and the divided areas are set for each inspection start point. It is characterized by being.

  The invention according to claim 10 is a visual inspection method, wherein a first step of setting an inspection start point on a boundary line of an inspection region set for an edge image generated from an original image which is a gray image of an inspection object; First, a detection area set linearly and at a constant width passing through the inspection start point is rotated around the inspection start point in the inspection area, and a pixel group that is a crack candidate is extracted at each position of the detection area. 2 steps, a third step of generating a divided area in which the inspection area is divided by a dividing line composed of a plurality of arcs of different diameters centered on the inspection start point, and a predetermined number or more of the plurality of divided areas And a fourth step of determining the crack candidate as a crack when there is a crack candidate that spans. To.

  The invention according to claim 11 is an appearance inspection apparatus, which is an imaging unit that captures a gray image of an object to be inspected, an edge image generation unit that generates an edge image from the gray image obtained by the imaging unit, An inspection area setting section for setting an inspection area, an inspection start point setting section for setting an inspection start point on the boundary line of the inspection area, and a detection area having a linear and constant width passing through the inspection start point. Rotating and scanning the detection area within the range of the inspection area with the center of the center, and a processing unit that extracts pixel groups that are crack candidates at each position of the detection area, and the inspection area is divided into a plurality of small sections through the inspection start point Set the subdivision lines to be performed and the subdivisions that pass through multiple subdivision points on the subdivision lines to make each subdivision in different directions. A divided area generating unit that generates a divided divided area, and a crack determining unit that determines a crack candidate as a crack when there are crack candidates extending over a specified number of divided areas. To do.

  According to the invention of claim 1, when checking whether or not it is a crack by verification of continuity for a crack candidate detected by rotating and scanning a narrow detection area around the inspection start point, Set the auxiliary dividing line that divides the inspection area into a plurality of subdivisions through the inspection start point, and divide each subdivision in different directions by dividing lines passing through the plurality of dividing points on the auxiliary dividing line When a crack candidate spans more than a specified number of divided areas, the crack candidate is determined to be a crack, so the divided area to be set to confirm whether the crack candidate is a crack should be set empirically. Therefore, it is possible to set automatically, and labor saving of the setting work of the divided area becomes possible. In addition, since the divided areas can be automatically generated according to a certain rule, the cracks can be detected with high reproducibility without variation in the accuracy of determining whether or not the crack is a crack. Further, since the number of divisions of the divided areas becomes almost the same regardless of the extension direction of the crack candidates, it is possible to verify the continuity of the cracks with almost the same accuracy regardless of the extension direction of the crack candidates.

  According to the invention of claim 2, since a part of the polygonal outline centered on the inspection start point is used as the dividing line, the directionality is compared with the case where a part of the arc is used as the dividing line. It is easy to determine what occurs, and processing with less load is possible.

  According to the invention of claim 3, since the inspection area is divided into two small sections by the auxiliary dividing line in the 45 degree direction, and the divided areas are set differently for each small section, the crack Regardless of the extension direction of the candidates, the continuity of the cracks can be determined with substantially the same accuracy.

  According to the invention of claim 4, since the inspection area is divided into two small sections by the auxiliary dividing line that is a diagonal line, and the divided areas are set differently for each small section, the crack candidate Regardless of the extending direction, the continuity of cracks can be determined with substantially the same accuracy. In addition, since the diagonal line is used as the auxiliary dividing line, setting of the auxiliary dividing line is easy.

  According to the invention of claim 5, since the inspection area is divided into three subdivisions by two auxiliary division lines, and the division areas are set to have different division directions for each subdivision, the crack candidate Regardless of the extending direction, the continuity of cracks can be determined with substantially the same accuracy. Moreover, by dividing into three small sections, it is possible to accurately determine the continuity of the crack extended from the vicinity of the inspection start point on one side of the inspection area.

  In the inventions of claims 3 to 5, since the inspection area is rectangular and the inspection area is divided into two or three small sections by a linear auxiliary dividing line, the determination of the continuity of cracks is made. It is easy and enables high-speed determination with a small processing load.

  According to the invention of claim 6, since the dividing point on the auxiliary dividing line is defined by a plurality of concentric circles centered on the inspection start point, the dividing line can be set geometrically, Variation in inspection accuracy can be prevented.

  According to the invention of claim 7, since the number of divisions is set for each small section, the number of divisions is defined so that the accuracy of verifying the continuity of cracks is constant regardless of the extension direction of the crack candidates. It becomes possible. Alternatively, the number of divisions can be set so that the accuracy of verifying the continuity of the cracks varies depending on the extension direction of the crack candidates.

  According to the invention of claim 8, since the range of the rotation operation of the detection area can be limited, the possibility of erroneous detection by detecting a crack candidate in an unnecessary range can be reduced.

  According to the ninth aspect of the present invention, it is possible to inspect a plurality of cracks in one inspection region by setting a plurality of inspection start points in the inspection region.

According to the invention of claim 10,
For crack candidates detected by rotating and scanning a narrow detection area around the inspection start point, when checking whether or not it is a crack by continuity verification, the diameter around the inspection start point is Dividing the inspection area into a plurality of divided areas using a plurality of different arcs as dividing lines, and when a crack candidate extends over a predetermined number of divided areas, the crack candidate is determined to be a crack. Therefore, it is possible to automatically set the divided area to be set to confirm whether or not it is empirically set, and to save labor for setting the divided area. In addition, since the divided areas can be automatically generated according to a certain rule, the cracks can be detected with high reproducibility without variation in the accuracy of determining whether or not the crack is a crack. In addition, since the dividing line is a part of an arc centered on the inspection start point, the continuity of the crack can be detected in the inspection area without directionality. Further, since the number of divisions of the divided areas becomes almost the same regardless of the extension direction of the crack candidates, it is possible to verify the continuity of the cracks with almost the same accuracy regardless of the extension direction of the crack candidates.

  According to the structure of the invention of claim 11, when confirming whether or not the crack candidate is detected by rotating and scanning the narrow detection area around the inspection start point, it is a crack by verifying continuity. In addition, an auxiliary dividing line that divides the inspection area into a plurality of small sections through the inspection start point is set, and each small section is divided in different directions by dividing lines that pass through the plurality of dividing points on the auxiliary dividing line. When a split area is generated, and the crack candidates cross over a predetermined number of split areas, the crack candidate is determined to be a crack, so experience the split area set to check whether the crack candidate is a crack or not. Therefore, it is possible to set automatically without setting, so that the labor for setting the divided area can be saved. In addition, since the divided areas can be automatically generated according to a certain rule, there is no variation in the accuracy of determining whether or not a crack is present, and a crack can be detected with good reproducibility. Moreover, since the number of divisions of the divided areas becomes almost equal regardless of the extension direction of the crack candidates, it is possible to verify the continuity of the cracks with almost the same accuracy regardless of the extension direction of the crack candidates.

It is a block diagram which shows embodiment. It is a figure which shows the concept of the inspection start point in the same as the above. It is a figure which shows the concept of the detection area in the same as the above. It is a figure which shows the example of a setting of the division area in the same as the above. It is a figure which shows another example of a setting of the division area in the same as the above. It is a figure which shows the setting method of the dividing line in the same as the above. It is operation | movement explanatory drawing same as the above. It is a figure which shows the other example of a setting of the divided area in the same as the above. It is a figure which shows the further another example of a setting of the division area in the same as the above. It is operation | movement explanatory drawing same as the above.

  In the present invention, a technique for distinguishing between a crack generated in an inspection object such as a synthetic resin molded product and a foreign substance such as dust attached to the surface of the inspection object will be described. Since the foreign matter adhering to the object to be inspected can be removed by wiping, the so-called waste splash can be prevented by treating the inspected object to which the foreign object is attached as a non-defective product. That is, it is possible to suppress a decrease in yield due to waste splashing by distinguishing between an inspection object to which foreign matter is attached and an inspection object having a crack.

  As shown in FIG. 1, a TV camera 2 is used as an imaging unit that images a spatial region including the inspection object 1. An original image captured by the TV camera 2 is input to the image processing apparatus 10. The image processing apparatus 10 is realized by executing an appropriate program by a computer. Therefore, although not shown in FIG. 1, at least a keyboard and a mouse can be used as the operation input device, and a CRT or a liquid crystal display can be used as the monitor device. The operation input device is connected to the input interface 21, and the monitor device is connected to the image output unit 22.

  An image captured by the TV camera 2 is captured through an image capturing unit 23 including a video capture board attached to the computer, and is stored in the image storage unit 11 as a digital image. As the image storage unit 11, a semiconductor memory is usually used, but a mass storage device such as a hard disk device may be used in combination as required.

  The original image is a grayscale image, and the edge image is generated by subjecting the original image stored in the image storage unit 11 to known differentiation or thinning in the edge image generation unit 12. The edge image is stored in the image storage unit 11. When generating an edge image, a differential absolute value (differential value) corresponding to the magnitude of the density gradient is obtained, and a direction orthogonal to the density gradient direction is a differential direction value (for each angle or direction with respect to the reference direction). Use the assigned code). The differential direction value corresponds to the extending direction of the edge.

  In the present embodiment, the differential absolute value and the differential direction value are obtained using the density of a 3 × 3 pixel square region (in other words, in the vicinity of 8) centered on the pixel of interest. The differential absolute value is the difference between the sum of the densities of three pixels on each side along the horizontal direction in the square area and the sum of the density of three pixels on each side along the vertical direction in the square area. It can obtain | require as ratio with the difference between. Further, as the differential direction value, a value obtained by rotating the angle obtained by the arctangent of the differential absolute value by 90 degrees and assigning codes of 0 to 7 to 8 angle sections of 45 degrees each is used. For example, -22.5 to 22.5 degrees is set to 0, and 157.5 to 202.5 degrees is set to 4. The above-described method for obtaining the differential absolute value and the differential direction value is an example, and other known methods can be used.

  By the way, since the edge image only needs to be obtained for the region of interest of the inspection object 1, an inspection region setting unit 13 is provided to limit the region. The inspection area setting unit 13 sets an inspection area of interest in the original image or the edge image. The inspection area is set within a range of one surface facing the TV camera 2 in the inspection object 1. The shape of the inspection area is not particularly limited, and a circular or elliptical inspection area can be set according to the shape to be detected. In the following description, a rectangular shape that is easy to set is used. The inspection area is used.

  The inspection area setting unit 13 manually sets an inspection area using an input device while confirming an original image or an edge image on a screen of a monitor device as an output device. Since the position of the inspection object 1 with respect to the TV camera 2 is normally controlled by a separate device, the inspection area can be fixed at a fixed position in the image after the inspection area is once set. However, when this configuration is adopted, it is necessary to periodically correct the inspection area. The inspection area setting unit 13 may employ a configuration in which an inspection area is automatically extracted by image processing using a change in density of the original image.

  In the following, an original image obtained by imaging the inspection object 1 is stored in the image storage unit 11, an inspection region is specified in the original image, and an edge image generated using the density in the inspection region is stored in the image storage unit 11. Assume that it is stored. In addition, a differential absolute value image having a differential absolute value used for generating an edge image as a pixel value and a differential direction value image having a differential direction value as a pixel value are stored in the image storage unit 11 in the inspection region. It shall be.

  As described above, the crack is longer than the foreign object, and the extension direction of the crack is almost determined according to the part of the inspection object 1, so if the information on the extension direction and the length is extracted, the crack and the foreign object Can be distinguished. Below, the technique which identifies both using the feature of the shape of a foreign substance like a crack and dust is demonstrated.

  For example, the inspection object 1 is assumed to be a wiring device including a container of a synthetic resin molded product, and the container is formed by joining two members (referred to as a body and a cover). That is, an assembly pawl is projected on the outer surface of the body, and a rectangular assembly hole provided on an assembly leg extending from the cover is engaged with the assembly pawl so that the body and the cover are coupled. The wiring device is the inspection object 1. In the wiring device having this configuration, since the body and the cover are coupled using the elasticity of the assembly leg, a crack may occur when the assembly leg is deformed. In particular, since the assembly hole is rectangular, there is a high possibility that a crack will occur near the corner of the assembly hole.

  In the present embodiment, as shown in FIG. 2, the inspection area Dd is rectangular, the long side of the inspection area Dd is made to match the horizontal direction of the original image, and the short side is made to match the vertical direction of the original image.

  When the inspection area Dd is set, the inspection start point setting unit 14 sets a plurality of inspection start points Ps on the boundary line Lb surrounding the inspection area Dd. The inspection start point setting unit 14 sets the inspection start point Ps at a predetermined position with respect to the inspection region Dd set by the inspection region setting unit 13. The inspection start point Ps can be set at a plurality of positions shown in FIG. However, in the embodiment described below, only the case where the inspection start point Ps is set to the upper left corner of the inspection region Dd and the case where it is set to the left side of the inspection region Dd is illustrated.

  When the inspection start point Ps is set, as shown in FIG. 3, the processing unit 15 sets a detection area Ec having a straight line and a constant width passing through the inspection start point Ps, and the inspection start point Ps is set. The detection area Ec is scanned around. However, the scanning range of the detection area Ec is limited to the inspection area Dd.

  Hereinafter, the process of scanning the detection area Ec around the inspection start point Ps is referred to as rotational scanning. At the time of rotational scanning of the detection area Ec, for example, the position of the detection area Ec is changed in increments of 1 degree. Here, since the detection area Ec has a scanning range only in the inspection area Dd, the length of the detection area Ec expands and contracts according to the rotation position.

  In the example of FIG. 3, when the detection area Ec is in contact with the long side of the inspection area Dd, the length of the detection area Ec matches the length of the long side of the inspection area Dd, and the detection area Ec is in the inspection area Dd. The length of the detection area Ec matches the length of the short side of the inspection area Dd.

  In the illustrated example, the range in which the detection area Ec is rotationally scanned is 90 degrees, and the length of the detection area Ec is maximized when the detection area Ec overlaps the diagonal line of the inspection region Dd. The detection area Ec is rotated and scanned clockwise in the inspection area Dd. Therefore, in the example shown in FIG. 3, the rotational scanning is started from the position where the detection area Ec overlaps the long side of the inspection region Dd, and the rotational scanning is ended at the position overlapping the short side.

  The above-described operation assumes that the detection area Ec is rotationally scanned over the entire inspection area Dd. However, by limiting the angle range in which the detection area Ec is rotationally scanned, erroneous detection can be prevented. In addition, the processing time associated with rotational scanning is reduced. The rotational scanning angle range can be set by the rotational scanning angle setting unit 16. That is, when an angular range of rotational scanning is instructed by the scanning input device connected to the input interface 21, the rotational scanning angular range is set in the rotational scanning angle setting unit 16, and the processing unit 15 detects the detection area Ec in the angular range. Rotational scanning is performed.

  The width of the detection area Ec is empirically set so as to be slightly larger than the width of the crack. That is, the width of the detection area Ec is set so that most of the crack is included in the detection area Ec when the detection area Ec overlaps the crack. However, if the width of the detection area Ec is too large, a single detection area Ec may include a plurality of cracks. Therefore, the width of the detection area Ec is set to include only one crack.

  During the rotational scanning of the detection area Ec, raster scanning is performed in the detection area Ec for each position of the detection area Ec, and pixels whose differential absolute value exceeds the specified value are extracted. That is, the presence / absence of a pixel present on the edge of a defective part such as a crack is searched. The prescribed value for the differential absolute value is determined empirically so that an edge corresponding to a crack can be detected.

  When a pixel whose differential absolute value exceeds a specified value exists in the detection area Ec, the differential direction value is compared with a specified angle range for the pixel. That is, an angular range corresponding to the extending direction of the crack for each position (angle) of the detection area Ec (in other words, a predetermined angular range with respect to the extending direction of the detection area Ec) during the rotational scanning of the detection area Ec. Is determined in advance, and when the differential direction value is obtained as described above, if the differential direction value is within this angular range, the pixel is stored as a defect candidate point. The defect candidate point is detected by the processing unit 15. When defect candidate points are extracted, the processing unit 15 integrates the differential absolute values of the defect candidate points.

  The processing unit 15 obtains the total number of defect candidate points and the sum of differential absolute values of defect candidate points for each position of the detection area Ec, and both the total number of defect candidate points and the sum of differential absolute values of defect candidate points are obtained. When the threshold values defined for the respective areas are exceeded, the connection area of the defect candidate points in the detection area Ec is determined as a crack candidate.

  In this case, since the narrow detection area Ec is set, the foreign matter and the crack attached to the inspection object 1 in the inspection region Dd can be separated. It is possible to discriminate between a crack and a foreign object by inspecting the feature amount.

  By the way, in the above-described processing, crack candidates are detected by comparing the total number of defect candidate points and the sum of differential absolute values related to defect candidate points with respective threshold values. In order to verify that it is a crack, the continuity of crack candidates is also confirmed.

  In this embodiment, in order to confirm the continuity of the crack candidates, the inspection area Dd is divided into a plurality of divided areas, and a group of pixels (connected areas) extracted as crack candidates extends over the plurality of divided areas. Adopt a method of determining whether or not it exists. The present embodiment has a feature in a technique for setting divided areas so that the continuity of cracks can be confirmed regardless of the extension direction of the cracks.

  That is, when the inspection start point Ps is set at one end of the crack as an ideal state, the crack is extended from the inspection start point Ps in the inspection region Dd. In other words, it can be said that the extending direction of the crack is substantially along the radial direction of the circle centering on the inspection start point Ps.

  Therefore, when a plurality of arcs having different diameters with the inspection start point Ps as the center are set, and the division area is formed by dividing the inspection area Dd by the division lines using the arcs as the division lines, the cracks extend in the extending direction. Therefore, there is a high possibility that the crack intersects the dividing line. That is, a plurality of concentric circles with the inspection start point Ps as the center are set, and a portion included in the inspection region Dd is used as a dividing line among the outlines of the concentric circles, and the inspection region Dd is divided into a plurality of dividing areas by the dividing lines. It is divided. Each dividing line is desirably set at equal intervals.

  If a plurality of divided areas as described above are formed, cracks are expected to be detected across a plurality of divided areas. Therefore, a threshold is set for the number of divided areas through which crack candidates pass. By prescribing, it is possible to confirm the continuity of crack candidates and to determine whether or not the crack is a crack. That is, the divided area generation unit 17 sets divided areas, and the crack determination unit 18 compares the number of divided areas where the crack candidates are straddled with a threshold value. If the threshold value is exceeded, it is determined that a crack exists. is there.

  By the way, as described above, when a divided area is formed by using a dividing line that is a part of an arc, the processing load for comparing the coordinate position of the dividing line with the coordinate position of a pixel included in a crack candidate increases. Therefore, as shown in FIG. 4B, a part of the polygonal outline may be used as the dividing line Ld.

  When setting such a dividing line Ld, it is desirable to set the vertex position of each polygon so that the sides of each polygon are parallel to each other. When setting such a dividing line Ld, as shown in FIG. 4A, the inspection region Dd is divided into a plurality of small sections Bs by a linear auxiliary dividing line La passing through the inspection start point Ps. As shown in FIG. 4B, a plurality of partial dividing lines Lp that intersect with the auxiliary dividing line La and are parallel to each other are set for each of the small sections Bs. In the illustrated example, it is divided into five small sections Bs by four auxiliary dividing lines La.

  Here, the dividing line which connected the partial dividing line Lp of the adjacent small division Bs by setting a position so that a pair of partial dividing line Lp of the adjacent small division Bs has an intersection on the auxiliary dividing line La. Ld can be formed. That is, a plurality of divided areas Ed having a part of the polygonal outline as the dividing line Ld are formed in the inspection region Dd.

  When the divided area Ed obtained by dividing the inspection area Dd by the dividing line Ld shown in FIG. 4B is used, as in the case where the inspection area Dd is divided by the concentric dividing line Ld, the divided area spans the crack candidates. The continuity of cracks can be confirmed by comparing the number of Ed with a threshold value.

  The dividing line Ld as shown in FIG. 4B using a part of the polygonal outline is less in processing load than the case where a part of the arc is used as the dividing line Ld. Since the processing load increases as the number increases, the processing load is still relatively large.

  Here, in order to reduce the processing load when a part of the polygonal outline is used as the dividing line Ld, the partial dividing line Lp is set to be parallel to the horizontal direction or the vertical direction of the edge image. Is considered effective. If the partial dividing line Lp is set in the horizontal direction, the continuity of the crack candidates can be confirmed by paying attention only to the vertical coordinate position, and the partial dividing line Lp is set in the vertical direction. This is because the continuity of crack candidates can be confirmed by paying attention only to the coordinate position in the horizontal direction.

  Therefore, when the inspection start point Ps is set at the corner of the inspection region Dd, a straight line that passes through the inspection start point Ps and divides the inspection region Dd into two small sections Bs as shown in FIG. As shown in FIG. 5 (b), a divided area Ed is formed by extending a partial dividing line Lp in a horizontal direction and a vertical direction from a dividing point where the auxiliary dividing line La is divided into a plurality of parts. Form. That is, the intersection of the partial dividing lines Lp in each small section Bs coincides with the dividing point on the auxiliary dividing line La.

  To set a dividing point on the auxiliary dividing line La, as shown in FIG. 6, a plurality of arcs Cc centering on the inspection start point Ps are set at equal intervals, and each arc Cc and the auxiliary dividing line La are set. The intersection point with may be used as a dividing point.

  Now, as shown in FIG. 7A, a case is assumed in which cracks C1 and C2 are generated from the vicinity of the upper left corner of the object 1 to be inspected. Here, as shown in FIG. 7B, a rectangular inspection region Dd is divided by a vertical dividing line Ld ′, and a plurality of divided areas Ed ′ are provided in the horizontal direction, and FIG. As shown in (), the case where the divided area Ed is provided using the L-shaped dividing line Ld with one diagonal line of the rectangular inspection region Dd as the auxiliary dividing line La will be compared.

  If the threshold value for confirming the continuity of the crack candidates is 5, if the strip-shaped divided area Ed ′ as shown in FIG. 7B is used, the number of cracks C1 extending along the horizontal direction is five. It is determined that there is continuity by straddling the divided area Ed ′, and is detected as a crack. On the other hand, since the crack C2 extending along the vertical direction extends over only the three divided areas Ed ′, it is determined that there is no continuity.

  On the other hand, when an L-shaped divided area Ed as shown in FIG. 7C is used, a crack C1 extending along the horizontal direction and a crack C2 extending along the vertical direction are formed. Since both span five divided areas Ed, it is determined that all have continuity.

  That is, by using the L-shaped divided area Ed, it is possible to confirm the continuity of the crack regardless of the extending direction of the cracks C1 and C2.

  In the illustrated example, the diagonal line of the inspection region Dd is used as the auxiliary dividing line La, but a straight line passing through the inspection start point Ps and forming 45 degrees with respect to the horizontal direction can also be used as the auxiliary dividing line La. The former auxiliary dividing line La is easy to set, and in the latter auxiliary dividing line La, the auxiliary dividing lines La are equally spaced in each of the small sections Bs. Can be confirmed.

  In the case where the diagonal line of the inspection region Dd is used as the auxiliary dividing line La as in the former case, in order to make the accuracy of continuity confirmation equal regardless of the extension direction of the crack candidates, as shown in FIG. The partial dividing lines Lp may be set at equal intervals for each small section Bs. In this case, the partial dividing line Lp of each small section Bs does not intersect on the auxiliary dividing line La, but the continuity of crack candidates can be confirmed with the same accuracy in each small section Bs.

  In the above example, the case where the inspection start point Ps is set at the corner of the inspection region Dd has been described. However, the inspection start point Ps is set at one side of the inspection region Dd as shown in FIG. There is also. Setting the inspection start point Ps to one side of the inspection region Dd is equivalent to connecting two inspection regions Dd having the inspection start point Ps set at the corner, and therefore the inspection start point Ps at the corner. And the dividing line Ld may be set.

  In the example shown in FIG. 9, a linear auxiliary dividing line La passing through the inspection start point Ps is set in the same manner as when the diagonal line of the inspection region Dd is used as the auxiliary dividing line La. Here, if the inspection area Dd shown in FIG. 9 is divided vertically by a horizontal straight line passing through the inspection start point Ps, it can be considered that two inspection areas having the inspection start point Ps connected to the corner are connected. By using a diagonal line passing through the inspection start point Ps for each connected inspection region as the auxiliary dividing line La, the dividing line Ld can be set in the same procedure as in the example shown in FIG.

  In this example, as shown in FIG. 9A, three auxiliary sections Bs are formed by setting two auxiliary dividing lines La in the inspection area Dd, and each of the small sections Bs is arranged in the horizontal direction or the vertical direction. The divided area Ed can be formed by dividing by the partial dividing line Lp. Similar to the above-described example, the partial dividing line Lp may be set so that the intersection of the partial dividing lines Lp in the adjacent small section Bs is positioned on the auxiliary dividing line La, but it is set to the adjacent small section Bs. It is not essential for the partial dividing line Lp to have an intersection point on the auxiliary dividing line La, and the intervals of the partial dividing lines Lp in the respective small sections Bs may be equalized.

  The interval between the dividing lines Ld described above is set according to the inspection object 1, for example, an interval of 1 to 2 mm. Further, the number of divided areas Ed formed in the inspection region Dd is arbitrarily set, for example, with an upper limit of 16. The threshold value for confirming the continuity of the crack candidates and determining as a crack may be set to 5 to 8, for example. That is, when it is confirmed that the crack candidate extends over 5 to 6 divided areas Ed, the crack candidate is determined as a crack.

  FIG. 10 shows a summary of the processing procedure for the crack detection technique described above. First, the inspection object 1 is imaged by the TV camera 2 (S1), and an edge image is generated using the grayscale image of the inspection object 1 (S2). The inspection region setting unit 13 sets the inspection region Dd in the edge image, and the inspection start point setting unit 6 sets the inspection start point Ps for the inspection region Dd. Further, the processing unit 15 sets a detection area Ec and performs rotation scanning of the detection area Ec (S4). While performing rotation scanning of the detection area Ec, a crack candidate is detected using the differential absolute value and the differential direction value (S3).

  When a crack candidate is detected, a divided area Ed obtained by dividing the inspection region Dd into a plurality of parts is set by the divided area generation unit 17, and the continuity of the crack candidates is confirmed in the crack determination unit 18 (S5). If the number of the divided areas Ec over which the crack candidates are spread is equal to or greater than the threshold (that is, if there is continuity), the crack candidates are determined as cracks (S6). On the other hand, if there is no crack candidate (S3: NO), or if there is no continuity in the crack candidate (S5: NO), it is determined that the crack is a good product (S7).

  In the above example, the description has been given focusing on one inspection start point Ps in the inspection area Dd. However, as described above, since a plurality of inspection start points Ps are set in the inspection area Dd, each inspection start point Ps is set. By setting the detection area Ec for each point Ps, it becomes possible to inspect a plurality of places using one inspection region Dd.

DESCRIPTION OF SYMBOLS 1 Inspection object 2 TV camera 10 Image processing apparatus 11 Image storage part 12 Edge image generation part 13 Inspection area setting part 14 Inspection start point setting part 15 Processing part 16 Rotation scanning angle setting part 17 Divided area generation part 18 Crack determination part Bs Subdivision C1, C2 Crack Dd Inspection area Ec Detection area Ed Divided area La Auxiliary dividing line Ld Dividing line Lp Partial dividing line Ps Inspection start point

Claims (11)

  The first step of setting the inspection start point on the boundary line of the inspection region set for the edge image generated from the original image which is the gray image of the inspection object, and the linear shape and the constant width passing through the inspection start point are set. A second step of rotating the detection area around the inspection start point in the inspection area and extracting a pixel group as a crack candidate at each position of the detection area, and a plurality of inspection areas through the inspection start point A third step of setting a subdivision line to be divided into a plurality of subdivisions and generating a subdivision area in which each subdivision is divided in different directions by a division line passing through a plurality of division points on the subdivision line, and a plurality of divisions When there are crack candidates that extend beyond the specified number of areas, the crack candidate is judged as a crack. Fourth appearance inspection method characterized by comprising the steps of:.   The appearance inspection method according to claim 1, wherein the dividing line is a part of a polygonal outline centered on the inspection start point.   The inspection area is rectangular, and the inspection start point is set at one vertex of the inspection area. In the third step, the inspection area passes through the inspection start point and intersects the vertical axis of the edge image at 45 degrees. The inspection area is divided into two subdivisions by the auxiliary dividing line, and the inspection area is divided into a plurality of inspection areas by perpendicular lines respectively dropped on two sides of the inspection area sandwiching the inspection start point from a plurality of dividing points set on the auxiliary dividing line. The appearance inspection method according to claim 1, wherein the visual inspection is divided into divided areas.   The inspection area is rectangular, and the inspection start point is set at one vertex of the inspection area, and in the third step, the diagonal line of the inspection area passing through the inspection start point is used as the auxiliary dividing line for inspection with an auxiliary dividing line. The area is divided into two subdivisions, and the inspection area is divided into a plurality of divided areas by perpendicular lines respectively extending from the plurality of division points set on the auxiliary dividing line to the two sides of the inspection area sandwiching the inspection start point. The appearance inspection method according to claim 1.   The inspection area is rectangular and the inspection start point is set on one side of the inspection area. In the third step, two straight lines that pass through the inspection start point and connect the opposite end points of the inspection area are respectively connected. The inspection area is divided into three subdivisions by two auxiliary dividing lines as the auxiliary dividing lines, and a perpendicular line dropped from a plurality of dividing points set in each auxiliary dividing line to the remaining two sides of the inspection area; 2. The appearance inspection method according to claim 1, wherein the inspection region is divided into a plurality of divided areas by a straight line connecting the dividing points on each auxiliary dividing line.   6. The appearance inspection method according to claim 3, wherein an intersection of a plurality of concentric circles centering on the inspection start point and the auxiliary dividing line is set as the dividing point.   The appearance inspection method according to claim 3, wherein the number of divisions is set for each of the small sections in the division area.   The visual inspection method according to claim 1, wherein, in the second step, an angle range for rotating the detection area is a range designated in the inspection region.   9. The method according to claim 1, wherein in the first step, a plurality of the inspection start points are set in the inspection area, and the divided area is set for each inspection start point. Visual inspection method.   The first step of setting the inspection start point on the boundary line of the inspection region set for the edge image generated from the original image which is the gray image of the inspection object, and the linear shape and the constant width passing through the inspection start point are set. A second step of rotating the detection area around the inspection start point within the inspection area and extracting a pixel group as a crack candidate at each position of the detection area, and a different diameter centered on the inspection start point A third step of generating a divided area in which the inspection area is divided by a dividing line composed of a plurality of arcs; and when there are crack candidates extending over a specified number of the divided areas, the crack candidates are defined as cracks. And a fourth step of determining the appearance inspection method.   Imaging means for capturing a grayscale image of an inspection object, an edge image generating section for generating an edge image from the grayscale image obtained by the imaging means, an inspection area setting section for setting an inspection area in the edge image, and an inspection area An inspection start point setting unit that sets the inspection start point on the boundary line of the inspection, and a linear and constant detection area that passes through the inspection start point is set, and the detection area is rotated around the inspection start point within the inspection area. And a processing unit for extracting a pixel group as a crack candidate at each position of the detection area, and an auxiliary dividing line that divides the inspection region into a plurality of small sections through the inspection start point and a plurality of the auxiliary dividing lines Generation of divided areas that generate divided areas by dividing each subdivision in different directions by dividing lines that pass through the dividing points If, appearance inspection apparatus; and a crack and determining crack determination unit the crack candidate when cracks candidates spanning more than the specified number of the plurality of divided areas are present.

Images