JP2014106198A - Defect determination device, correspondence information creation device, defect determination method, and correspondence information creation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a defect determination device capable of accurately determining a defect.SOLUTION: A defect determination device (11) includes: an attention pixel selection part (111) for selecting one of a plurality of pixels as one attention pixel; a reference pixel group extraction part (112) for extracting a reference pixel group as a pixel group constituted of a plurality of pixels in a predetermined range around the attention pixel, that is, the pixel group excluding the attention pixel and pixels adjacent to the attention pixel on the basis of predetermined setting condition information; a defect candidate determination value deriving part (113) for calculating a mean luminance value in the reference pixel group, and for deriving a defect candidate determination value corresponding to the mean luminance value on the basis of predetermined correspondence information; a differential value calculation part (114) for calculating a differential value between the luminance value of the attention pixel and the mean luminance value; and a defect candidate determination part (115) for determining whether or not the attention pixel is a defect candidate pixel on the basis of the differential value and the defect candidate determination value.

Description

  The present invention relates to a defect determination device, a correspondence relationship information creation device, a defect determination method, and a correspondence relationship information creation method.

  A sheet body such as a liquid crystal panel is treated as a defective product because its characteristics change if there are defects such as minute irregularities on the surface or inside. Therefore, it is necessary to inspect the presence or absence of defects before shipping the sheet body.

  Conventionally, the presence or absence of a defect has been inspected by photographing a sheet body to acquire image data, and determining an image based on the image data. Regarding image determination, Patent Documents 1 and 2 describe techniques for averaging the luminance values of a target pixel and peripheral pixels.

Japanese Patent No. 4893788 JP 2010-141435 A

  When the techniques described in Patent Documents 1 and 2 are applied to defect inspection, the difference value between the average value of the luminance values of the target pixel and the peripheral pixels and the luminance value of the target pixel is calculated, and the difference value is large In addition, it is possible to configure a defect determination apparatus that determines that the target pixel corresponds to a defect. It is considered that this defect determination device can accurately determine a defect if an illumination device for photographing is ideal.

  However, the illuminating device is rarely ideal, and if the accuracy of the arrangement position of the illuminating device is not good, or if the illuminating device has deteriorated over time, the luminance value non-uniformity unrelated to the defect, that is, , Noise will occur. Also, noise may occur depending on the shape and characteristics of the sheet body regardless of the presence or absence of defects. When there is not much difference between the luminance value of the pixel corresponding to the defect and the luminance value of the pixel in which noise occurs, the above-described defect determination device distinguishes the pixel corresponding to the defect from the pixel in which noise has occurred. As a result, the defect cannot be accurately determined.

  The present invention solves such a problem, and relates to a defect determination apparatus capable of accurately determining a defect, a correspondence information creation apparatus according to the defect determination apparatus, a defect determination method, and the defect determination method. The purpose is to provide a correspondence creation method.

The present invention is a defect determination device for determining whether or not each pixel of an image composed of a plurality of pixels representing a product to be inspected is a defect candidate pixel,
A pixel-of-interest selection unit that selects one of the plurality of pixels as a pixel of interest;
A reference pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the target pixel, the pixel group not including the target pixel and a pixel adjacent to the target pixel A reference pixel group extraction unit for extracting a pixel group;
Calculates the average luminance value of all the pixels constituting the reference pixel group, and corresponds to the calculated average luminance value based on predetermined correspondence information indicating the correspondence relationship between the defect candidate determination value and the average luminance value. A defect candidate determination value deriving unit for deriving a defect candidate determination value to be
A difference value calculation unit for calculating a difference value between the luminance value of the target pixel and the average luminance value;
And a defect candidate determination unit that determines whether or not the target pixel is a defect candidate pixel based on the difference value and the defect candidate determination value.

  In the present invention, it is preferable that the setting condition information includes that the predetermined range is a distance from the target pixel that is farther from the target pixel than a pixel that is adjacent to the target pixel. Is information in a range up to a distant second position.

Further, the present invention is a correspondence information creation device that creates the correspondence information in the defect determination device based on a model product selected in advance as a model of the inspection target product,
A model-target pixel selection unit that selects one of the plurality of pixels in an image including a plurality of pixels representing the model product as a target pixel;
A reference pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the target pixel, the pixel group not including the target pixel and a pixel adjacent to the target pixel A model reference pixel group extraction unit for extracting a pixel group;
An exemplary product average luminance value calculating unit for calculating an average luminance value of all the pixels constituting the reference pixel group;
An exemplary product difference value calculation unit for calculating a difference value between the luminance value of the target pixel and the average luminance value;
Based on the relationship between the average luminance value corresponding to each pixel of interest and the difference value, a defect candidate determination value corresponding to the average luminance value is calculated, and the correspondence relationship between the defect candidate determination value and the average luminance value A correspondence relationship information creating unit that creates correspondence relationship information indicating the correspondence relationship information.

The present invention is also a defect determination method executed by the information processing apparatus,
A pixel-of-interest selection step in which the information processing apparatus selects one of the plurality of pixels in an image composed of a plurality of pixels representing an inspection target product as a pixel of interest;
The information processing apparatus is a pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the pixel of interest, and includes the pixel of interest and a pixel adjacent to the pixel of interest A reference pixel group extraction step for extracting a reference pixel group that is not a pixel group;
The information processing apparatus calculates an average luminance value of all the pixels constituting the reference pixel group, and calculates based on correspondence information indicating a correspondence relationship between a defect candidate determination value and an average luminance value, which is determined in advance. A defect candidate determination value derivation step for deriving a defect candidate determination value corresponding to the average brightness value,
A difference value calculating step in which the information processing apparatus calculates a difference value between a luminance value of the target pixel and the average luminance value;
A defect determination method comprising: a defect candidate determination step for determining whether or not the target pixel is a defect candidate pixel based on the difference value and the defect candidate determination value. It is.

Further, the present invention is a correspondence information creation method executed by an information processing apparatus, which creates the correspondence information in the defect determination method based on a model product selected in advance as a model of the inspection target product. ,
The information processing device selects the one of the plurality of pixels in the image composed of a plurality of pixels representing the model product as a target pixel;
The information processing apparatus is a pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the pixel of interest, and includes the pixel of interest and a pixel adjacent to the pixel of interest A model reference pixel group extraction step for extracting a reference pixel group that is not a pixel group;
An exemplary product average luminance value calculating step in which the information processing apparatus calculates an average luminance value of all the pixels constituting the reference pixel group;
The information processing apparatus calculates a difference value between the luminance value of the target pixel and the average luminance value;
The information processing apparatus calculates a defect candidate determination value corresponding to the average luminance value based on a relationship between the average luminance value corresponding to each pixel of interest and the difference value, and the defect candidate determination value and the average A correspondence relationship information creating method comprising: a correspondence relationship information creating unit that creates correspondence relationship information indicating a correspondence relationship with a luminance value.

  According to the present invention, a reference pixel group including a plurality of pixels within a predetermined range around a target pixel in an image representing an inspection target product is extracted. Then, a defect candidate determination value corresponding to the average luminance value is derived from the average luminance value in the reference pixel group and predetermined correspondence information, and the defect candidate determination value, the luminance value of the target pixel, and the average luminance Whether or not the target pixel is a defect candidate pixel is determined based on a difference value from the value. In the prior art, when there is luminance unevenness in an image representing an inspection target product, pixels that are not defective are often erroneously determined as defect candidate pixels, particularly in a bright portion. On the other hand, in the present invention, whether or not the target pixel is a defect candidate pixel is determined based on the defect candidate determination value and the difference value between the luminance value and the average luminance value of the target pixel. Even if there is luminance unevenness in the image representing the inspection target product, it is possible to determine a pixel that is a defect as a defect candidate pixel with higher accuracy, and it is possible to accurately determine the defect.

  In addition, according to the present invention, a reference pixel group including a plurality of pixels within a predetermined range around a target pixel in an image representing a model product is extracted. Then, from the relationship between the average luminance value in the reference pixel group and the difference between the luminance value of the target pixel and the average luminance value, a defect candidate determination value corresponding to the average luminance value is calculated, and correspondence information is created. Is done. Therefore, even if noise occurs in any pixel in the reference pixel group, it is possible to create correspondence information that can accurately determine a defect. In addition, since the reference pixel group does not include the pixel of interest and the pixels adjacent to the pixel of interest, even if noise occurs in the pixel of interest and adjacent pixels, correspondence information that can accurately determine the defect is created. can do.

It is a perspective view showing inspection system 1 provided with defect judging device 11 concerning the present invention. It is the figure which looked at the inspection system 1 from the side. It is a block diagram showing inspection system 1 concerning the present invention. It is a flowchart which shows the correspondence information creation method which concerns on this invention. It is a figure for demonstrating the predetermined range based on a 1st position and a 2nd position. It is a graph which shows the relationship between each difference value and the average luminance value corresponding to each difference value. It is a figure for demonstrating the process of step S9. It is a flowchart which shows the defect determination method which concerns on this invention.

  FIG. 1 is a perspective view showing an inspection system 1 including a defect determination device 11 according to the present invention. FIG. 2 is a side view of the inspection system 1. FIG. 3 is a block diagram showing the inspection system 1 according to the present invention. The inspection system 1 is a system for inspecting whether or not there is a defect in the inspection target product using a sheet body such as a flat liquid crystal panel as the inspection target product.

  The sheet body to be inspected may have any thickness, may have a relatively thin thickness generally referred to as “film”, and is generally referred to as “plate”. It may have a relatively thick thickness. For example, examples of the sheet body include a transparent or translucent optical film made of a thermoplastic resin such as a polarizing film or a retardation film, and a transparent or translucent plate-like body such as a glass plate or a liquid crystal panel. . In addition, a sheet | seat body is not limited to a flat thing, You may have a gentle curvature.

  The sheet body as described above is treated as a defective product because the characteristics change if there are defects such as minute irregularities on the surface or inside. The inspection system 1 can determine whether the inspection target product is a non-defective product or a defective product by the defect determination device 11.

  As shown in FIGS. 1, 2, and 3, the inspection system 1 includes a conveyance unit 12 that conveys a sheet body in one longitudinal direction (hereinafter referred to as “Y direction”), and a width direction of the sheet body (hereinafter referred to as “Y direction”). , Referred to as “X direction”), a light irradiation unit 13 having a light source extending linearly, a photographing unit 14 for photographing a sheet, a conveyance control unit 15 for controlling the conveyance unit 12, and a defect determination device 11. Prepare.

  The defect determination apparatus 11 includes a target pixel selection unit 111, a reference pixel group extraction unit 112, a defect candidate determination value derivation unit 113, a difference value calculation unit 114, and a defect candidate determination unit 115. The defect determination apparatus 11 also includes a control unit (not shown) that controls the light irradiation unit 13, the imaging unit 14, and the conveyance control unit 15.

  The conveyance unit 12 is a device that conveys the sheet body in the Y direction, and moves in the Y direction along the rail portion 12a with the two rail portions 12a extending linearly in the Y direction and the sheet body placed thereon. Scanning stage 12b. The transport controller 15 controls the movement of the scan stage 12b. The moving speed of the scanning stage 12b, that is, the conveying speed of the sheet body is set to 100 mm / second to 500 mm / second, for example.

  The light irradiation unit 13 fixes the light source so that the linear light source extending in the X direction, which is a direction orthogonal to the Y direction, and the position of the light source become a fixed position with respect to the rail portion 12a of the transport unit 12 are not illustrated. A fixing member. The light source is disposed above the sheet body so as to face the surface of the sheet body. The light source is, for example, a linear light source in which LEDs (Light Emitting Diodes) are arranged in an array (or a line), a linear light source in which a halogen lamp and a linear optical fiber light guide are combined, a halogen lamp and a long length And a linear light source combined with the rod. The length of the light source in the Y direction is, for example, 30 mm, the length in the X direction is, for example, 400 mm, and the distance from the light source to the surface of the sheet member is, for example, 100 mm.

  The photographing unit 14 includes a line sensor camera of CCD (Charge Coupled Device) or CMOS (Complementary Metal-Oxide Semiconductor). The line sensor camera is accompanied by a lens or the like for forming an optical image of a sheet body to be photographed on a light receiving element. The line sensor camera is arranged above the sheet body so as to face the surface of the sheet body, and is focused so that the focal position exists on the surface of the sheet body, and the photographing position is fixed. The line sensor camera is photographing one line on the surface of the sheet body, and the entire image of the sheet body is photographed when the sheet body is conveyed in the Y direction.

  The defect determination device 11 includes a control arithmetic circuit such as a CPU (Central Processing Unit), a volatile memory such as a DDR SDRAM (Double Data Rate Synchronous Dynamic Random Access Memory), a flash ROM (Read Only Memory), and an EEPROM (registered trademark). ) And an information processing apparatus including a nonvolatile memory such as an HDD (Hard Disk Drive). In the nonvolatile memory, a program for causing the defect determination device 11 to function as the target pixel selection unit 111, the reference pixel group extraction unit 112, the defect candidate determination value derivation unit 113, the difference value calculation unit 114, and the defect candidate determination unit 115 Data is stored, and the defect determination apparatus 11 performs these functions according to the program data. Further, the nonvolatile memory includes an exemplary product attention pixel selection unit according to the present invention, an exemplary product reference pixel group extraction unit according to the present invention, an exemplary product average luminance value calculation unit according to the present invention, and an exemplary product difference according to the present invention. Program data for causing the defect determination device 11 to function as a value calculation unit and a correspondence relationship information creation unit according to the present invention is stored, and according to the program data, the defect determination device 11 performs correspondence relationship information according to the present invention. Functions as a creation device. The correspondence information creation device according to the present invention creates correspondence information by executing the correspondence information creation method according to the present invention, and the defect determination device 11 is based on the created correspondence information. Then, the defect determination method according to the present invention is executed to determine the presence or absence of defects.

  First, the correspondence information creation method according to the present invention will be described. FIG. 4 is a flowchart showing the correspondence information creation method according to the present invention.

  In step S1, a sheet of a model article prepared in advance and confirmed to be non-defective by human eyes is photographed by the photographing unit 14, and a model image that is a large image including an image representing the model article is acquired. Is done. The exemplary product image has a plurality of pixels arranged in the X direction and the Y direction.

  In step S <b> 2, the model attention area is set from the model image by the defect determination device 11. The model article attention area is an area corresponding to the model article in the model article image. The defect determination apparatus 11 detects the position of a predetermined alignment mark from the model product image, identifies the position of the model product based on the position, and sets the model product attention area.

  Step S3 is a model product attention pixel selection step according to the present invention, and in the model product attention pixel selection step, one of a plurality of pixels in the model product attention area set in step S2 is selected as a target pixel. It is selected by the model item attention pixel selection unit according to the present invention. Step S3 is repeated by step S7 to be described later, and all the pixels in the model item attention area are selected one by one in order. For example, the pixel at the left end of the top stage in the model product attention area is sequentially selected in the right direction, and when it goes to the right end, the next lower stage pixel is sequentially selected from the left end to the right end. Selection is sequentially made up to the lowest level in the region of interest.

  Step S4 is an exemplary product reference pixel group extraction step according to the present invention. In the exemplary product reference pixel group extraction step, a predetermined area around the target pixel selected in step S3 is determined based on predetermined setting condition information. A reference pixel group, which is a pixel group including a plurality of pixels within a range and does not include the pixel of interest and a pixel adjacent to the pixel of interest, is obtained by the exemplary product reference pixel group extraction unit according to the invention. Extracted. The setting condition information is information stored in the non-volatile memory of the defect determination device 11, and a predetermined range is set from the first position that is farther from the target pixel than the pixel adjacent to the target pixel. This is information that is a range up to a second position that is farther from the target pixel than the position.

  The predetermined range based on the first position and the second position will be described with reference to FIG. FIGS. 5A and 5B show a part of the model product attention area, pixels are arranged in the X direction and the Y direction, and among the arranged pixels, the hatched pixels are reference pixels. A group.

  In FIG. 5A, the pixel surrounded by the one-dot chain line is the pixel A1 at the first position, and the pixel surrounded by the two-dot chain line is the pixel B1 at the second position. The pixel A1 is at a position farther from the target pixel C1 than the pixel adjacent to the target pixel C1 in FIG. 5A, and the pixel B1 is at a position farther from the target pixel C1 than the pixel A1. is there. In FIG. 5A, pixels surrounded by the pixel A1 and the pixel B1 and the pixel A1 and the pixel B1 are pixels within a predetermined range, and all the pixels within the predetermined range are a reference pixel group.

  In FIG. 5B, the pixel surrounded by the one-dot chain line is the pixel A2 at the first position, and the pixel surrounded by the two-dot chain line is the pixel B2 at the second position. The pixel A2 is at a position farther from the target pixel C2 than the pixel adjacent to the target pixel C2 in FIG. 5B, and the pixel B2 is at a position farther from the target pixel C2 than the pixel A2. is there. In FIG. 5B, the pixels surrounded by the pixels A1 and B1 and the pixels A1 and B1 are pixels within a predetermined range, and among the pixels within the predetermined range, the target pixel C2 and the X coordinate value or Twelve pixels having the same Y coordinate value are a reference pixel group.

  The position of the reference pixel group with respect to the target pixels C1, C2 may be determined in advance as XY coordinate values corresponding to the X coordinate value and the Y coordinate value of the target pixels C1, C2, or the pixels A1, A2, B1, The XY coordinate value of B2 may be determined in advance, and the XY coordinate value of the reference pixel group may be calculated based on the XY coordinate value.

  The XY coordinate values of the pixels A <b> 1 and A <b> 2 are set according to the expected size of the defect that occurs in the sheet body that is the inspection target product. For example, in FIGS. 5A and 5B, the size of the defect in the image is expected to be 9 pixels × 9 pixels in the X direction and the Y direction. The XY coordinate value of A2 is set. The XY coordinate values of the pixels B1 and B2 are set such that, for example, the distance between the target pixels C1 and C2 and the pixels A1 and A2 is set so that the distance between the pixels B1 and B2 and the target pixels C1 and C2 is not too long. The distance is set to be about 1.5 to 3 times. In FIG. 5, the pixels A1, A2, B1, and B2 are arranged in a rectangular shape, but a circle centered on the target pixels C1 and C2 whose distance from the target pixels C1 and C2 is a predetermined number of pixels. It may be arranged in a shape.

  In step S4, when the reference pixel group is extracted as described above, the process proceeds to step S5. Step S5 is a model average brightness value calculation step according to the present invention. In the model average brightness value calculation step, the arithmetic average value of the brightness values in each pixel of the reference pixel group extracted in step S4, that is, the average The luminance value is calculated by the model average luminance value calculation unit according to the present invention. The luminance value is, for example, a value of 256 gradations, and is an integer value of 0 or more and 255 or less.

  Step S6 is a model product difference value calculation step according to the present invention. In the model product difference value calculation step, the difference value between the luminance value of the target pixel and the average luminance value corresponding to the target pixel is included in the present invention. It is calculated by the model item difference value calculation unit. When the luminance value is an integer value of 0 or more and 255 or less, the difference value is also an integer value of 0 or more and 255 or less.

  In step S <b> 7, the defect determination device 11 determines whether or not all the pixels in the model item attention area have been selected as the attention pixel. If all the pixels in the model attention area are selected as the attention pixel, the process proceeds to step S8. If any one of the pixels in the model attention area is not selected as the attention pixel, the process returns to step S3. .

  In step S8, a graph showing the relationship between each difference value calculated in step S6 and the average luminance value corresponding to each difference value is created. FIG. 6 shows this graph. In the graph of FIG. 6, the horizontal axis represents the average luminance value, and the vertical axis represents the difference value.

  Step S9 is a correspondence information creating step according to the present invention. In the correspondence information creating step, the correspondence information creating unit according to the present invention sets the average luminance value based on the graph created in step S8. Corresponding defect candidate determination values are calculated, and correspondence information indicating the correspondence between the defect candidate determination values and the average luminance value is created.

  The process of step S9 will be described using FIG. In step S9, first, a linear function is fitted to all points in the graph of FIG. 6 using the least square method. FIG. 7A shows the fitted linear function by a one-dot chain line. In step S9, the slope of this linear function is temporarily stored.

  Next, in step S9, the entire range of values that the average luminance value can take is divided into small ranges of 1/20 to 1/5 of the entire range. A broken line shown in FIG. 7B indicates each divided small range. In step S9, the point where the difference value is maximum is extracted in each small range. For example, in FIG. 7B, a point P at which the average luminance value is 55 and the difference value is 10 is a point where the difference value is maximum in a small range of the average luminance value of 47 to 63. In step S9, the coordinate value of each extracted point is temporarily stored.

  Next, in step S9, a line segment that passes through the extracted point and becomes the slope of the linear function fitted with the slope is set in each small range. Then, in each small range, the point on the set line segment, the average luminance value is the lower limit value or the upper limit value in the range, and the point larger than the point where the difference value is extracted The difference value is calculated. Each calculated difference value is a defect candidate determination value corresponding to each average luminance value in each small range. When no point is extracted in a certain small range, the line segment having the largest Y-intercept is selected from the line segments set in each small range, and the Y-intercept and the stored primary are selected. Based on a straight line determined by the slope of the function, an upper limit value in a small range without a point is set as a defect candidate determination value.

  In FIG. 7C, a line segment in each small range is indicated by a one-dot chain line, and a defect candidate determination value corresponding to each average luminance value in each small range is indicated by a two-dot chain line. Table 1 below shows an example of each average luminance value and a defect candidate determination value corresponding to each average luminance value. Table 1 is correspondence information according to the present invention, and is stored in the defect candidate determination value deriving unit 113.

  In addition, the calculation method of the defect candidate determination value in step S9 which is a correspondence information creation step according to the present invention is not limited to the above method. Other examples of the defect candidate determination value calculation method include the following method. That is, after determining the slope of the linear function shown in FIG. 7A, a point having the largest Y-intercept is extracted from all the points in the graph of FIG. 6, and defined by the Y-intercept and the slope. Determine the straight line to be played. Then, using the determined straight line, the average luminance value is substituted into the straight line expression without being divided into small ranges, and a defect candidate determination value is obtained as an output. In this way, the defect candidate determination value can also be calculated.

  When the defect candidate determination value is stored in the defect candidate determination value deriving unit 113 in this way, the defect determination device 11 uses the defect candidate determination value to determine the defect determination according to the present invention for the inspection target product. The method is executed to determine the presence or absence of defects. FIG. 8 is a flowchart showing a defect determination method according to the present invention.

  In step T1, a sheet body to be inspected is imaged by the imaging unit 14, and an inspection object image that is a large image including an image representing the inspection object is acquired. The product to be inspected is the same type as the model product. The product image to be inspected has a plurality of pixels arranged in the X direction and the Y direction.

  In step T <b> 2, the inspection target product attention area is set from the inspection target product image by the defect determination device 11. The inspection target product attention area is an area corresponding to the inspection target product in the inspection target product image. The defect determination device 11 detects the position of a predetermined alignment mark from the inspection object image, specifies the position of the inspection object based on the position, and sets the inspection object attention area.

  Step T3 is a target pixel selection step according to the present invention. In the target pixel selection step, one of the plurality of pixels in the inspection target product target region set in step T2 is selected as the target pixel. Selected by the unit 111. Step T3 is repeated by step T8 described later, and all the pixels in the inspection target product attention area are selected one by one in order. For example, the leftmost pixel in the inspection target product's attention area is sequentially selected in the right direction, and when going to the right edge, the next lower pixel is sequentially selected from the left edge to the right edge. The items are sequentially selected up to the lowest level in the target product attention area.

  Step T4 is a reference pixel group extraction step according to the present invention. In the reference pixel group extraction step, the reference pixel group extraction step is within a predetermined range around the target pixel selected in step T3 based on predetermined setting condition information. A reference pixel group that is a pixel group that includes a plurality of pixels and that does not include the pixel of interest and a pixel adjacent to the pixel of interest is extracted by the reference pixel group extraction unit 112. The setting condition information is information stored in the non-volatile memory of the defect determination device 11, and a predetermined range is set from the first position that is farther from the target pixel than the pixel adjacent to the target pixel. This is information that is a range up to a second position that is farther from the target pixel than the position. The setting condition information and the predetermined range have already been described with reference to FIG.

  When the reference pixel group is extracted in step T4, the process proceeds to step T5. Step T5 is a defect candidate determination value deriving step according to the present invention. In the defect candidate determination value deriving step, the defect candidate determination value deriving unit 113 calculates the average luminance value of the reference pixel group extracted in step T4. The defect candidate determination value corresponding to the average luminance value is derived by the defect candidate determination value deriving unit 113 from the calculated average luminance value and the correspondence relationship information stored in the defect candidate determination value deriving unit 113. Is done. For example, when the correspondence information in Table 1 is stored in the defect candidate determination value deriving unit 113 and the average luminance value of the reference pixel group is 85, the defect candidate determination value is derived as 17.

  Step T6 is a difference value calculation step according to the present invention. In the difference value calculation step, the difference value calculation unit 114 calculates a difference value between the luminance value of the target pixel and the average luminance value corresponding to the target pixel. Is done. This difference value is an absolute value obtained by subtracting the average luminance value from the luminance value of the target pixel. When the luminance value is an integer value of 0 or more and 255 or less, the difference value is also an integer value of 0 or more and 255 or less. .

  Step T7 is a defect candidate determination step according to the present invention, and the defect candidate determination step corresponds to the difference value based on the difference value calculated in step T6 and the defect candidate determination value derived in step T5. The defect candidate determination unit 115 determines whether the target pixel to be processed is a defect candidate pixel. Specifically, the defect candidate determination unit 115 compares the difference value with the defect candidate determination value, and when the difference value exceeds the defect candidate determination value, the target pixel corresponding to the difference value is a defect candidate pixel. The luminance value of the target pixel is replaced with “1”, and when the difference value is equal to or less than the defect candidate determination value, it is determined that the target pixel corresponding to the difference value is not a defect candidate pixel, and the target pixel The luminance value of the pixel is replaced with “0”. That is, the defect candidate determination unit 115 binarizes the luminance value of the target pixel based on the difference value and the defect candidate determination value.

  In step T8, the defect determination apparatus 11 determines whether all the pixels in the inspection target product attention area have been selected as the attention pixels. If all the pixels in the inspection target product attention area are selected as the attention pixels, the process proceeds to step T9. If any one of all the pixels in the inspection target article attention area is not selected as the attention pixel, step T3 is performed. Return to.

  In step T <b> 9, the defect determination apparatus 11 determines that the pixel having the luminance value “1”, that is, the defect candidate pixel in which no other defect candidate pixel is adjacent, and several other defect candidate pixels are adjacent. It is determined that the generated noise is randomly generated noise, and the noise is removed. Specifically, the defect determination apparatus 11 sequentially executes a contraction filter on each defect candidate pixel and assigns a temporary luminance value. The contraction filter sets the provisional luminance value to “0” if there is at least one pixel having a luminance value “0” among the eight pixels adjacent to the defect candidate pixel, and 1 pixel having the luminance value “0”. If not, the filter sets the temporary luminance value to “1”. When the defect determination apparatus 11 finishes assigning temporary luminance values to all defect candidate pixels, the defect determination device 11 replaces the luminance value “1” of the defect candidate pixels with the temporary luminance value. As a result, the noise is removed, and the defect candidate pixels remain only in the area where the defect candidate pixels are dense, that is, the defect candidate area.

  Since the area of the defect candidate region is reduced by the shrink filter, it is preferable to execute the expansion filter on each defect candidate pixel of the defect candidate region to increase the area of the defect candidate region. The expansion filter assigns a temporary luminance value “1” to the eight pixels if there is at least one pixel having a luminance value “1” among the eight pixels adjacent to the defect candidate pixel, and the luminance value “1”. If there is no pixel “”, the filter assigns a temporary luminance value “0” to the eight pixels. When the defect determination apparatus 11 finishes executing the dilation filter on each defect candidate pixel in all defect candidate areas, the pixel brightness value of the pixel assigned the temporary brightness value “1” at least once is set to “ For a pixel to which only the temporary luminance value “0” is assigned, the luminance value of the pixel is replaced with “0”.

  In step T10, the defect determination apparatus 11 attaches labels in order from 1 for each defect candidate area. Thereby, the defect determination apparatus 11 can recognize how many defect candidate areas are present.

  In step T11, the defect determination device 11 calculates feature amounts in the order of labels for each defect candidate region. The feature amount is, for example, the area (number of pixels) of the defect candidate region, the ferret diameter (X direction size, Y direction size) of the defect candidate region, the circularity of the defect candidate region, and the like.

  In step T12, the defect determination apparatus 11 determines whether each defect candidate area is a defect area based on predetermined defect determination information in the order of labels. The defect determination information is information stored in the non-volatile memory of the defect determination apparatus 11, for example, information that “a region having an area of 30 pixels or more is a defective region” or “an area is 12 pixels or more. And a region having a circularity of 0.8 or more is defined as a defect region.

  In step T13, the defect determination apparatus 11 determines whether the inspection target product is a non-defective product or a defective product based on the determination result in step T12 for each defect candidate region. For example, if there is even one defect determined in step T12 as a defective area, the defect determination apparatus 11 determines that the product to be inspected is a defective product, and one that is determined as a defective area in step T12. If not, it is determined that the product to be inspected is a non-defective product.

  According to the defect determination apparatus 11 as described above, a reference pixel group including a plurality of pixels within a predetermined range around a target pixel in an image representing an inspection target product is extracted. Then, a defect candidate determination value corresponding to the average luminance value is derived from the average luminance value in the reference pixel group and predetermined correspondence information, and the defect candidate determination value, the luminance value of the target pixel, and the average luminance Whether or not the target pixel is a defect candidate pixel is determined based on a difference value from the value. In the prior art, when there is luminance unevenness in an image representing an inspection target product, pixels that are not defective are often erroneously determined as defect candidate pixels, particularly in a bright portion. On the other hand, the defect determination apparatus 11 determines whether or not the target pixel is a defect candidate pixel based on the defect candidate determination value and the difference value between the luminance value and the average luminance value of the target pixel. Therefore, even when there is luminance unevenness in the image representing the inspection target product, it is possible to determine a defective pixel as a defective candidate pixel with higher accuracy, and to accurately determine the defect.

  Further, the defect determination device 11 functions as a correspondence relationship information creation device according to the present invention, thereby extracting a reference pixel group composed of a plurality of pixels within a predetermined range around a pixel of interest in an image representing an exemplary product. Is done. Then, from the relationship between the average luminance value in the reference pixel group and the difference between the luminance value of the target pixel and the average luminance value, a defect candidate determination value corresponding to the average luminance value is calculated, and correspondence information is created. Is done. Therefore, even if noise occurs in any pixel in the reference pixel group, it is possible to create correspondence information that can accurately determine a defect. In addition, since the reference pixel group does not include the pixel of interest and the pixels adjacent to the pixel of interest, even if noise occurs in the pixel of interest and adjacent pixels, correspondence information that can accurately determine the defect is created. can do. Note that an information processing apparatus other than the defect determination apparatus 11 may create the correspondence information.

DESCRIPTION OF SYMBOLS 1 Inspection system 11 Defect determination apparatus 12 Conveyance part 13 Light irradiation part 14 Imaging | photography part 111 Interesting pixel selection part 112 Reference pixel group extraction part 113 Defect candidate determination value deriving part 114 Difference value calculation part 115 Defect candidate determination part

Claims (5)

A defect determination device that determines whether each pixel of an image composed of a plurality of pixels that represents an inspection object is a defect candidate pixel individually,
A pixel-of-interest selection unit that selects one of the plurality of pixels as a pixel of interest;
A reference pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the target pixel, the pixel group not including the target pixel and a pixel adjacent to the target pixel A reference pixel group extraction unit for extracting a pixel group;
Calculates the average luminance value of all the pixels constituting the reference pixel group, and corresponds to the calculated average luminance value based on predetermined correspondence information indicating the correspondence relationship between the defect candidate determination value and the average luminance value. A defect candidate determination value deriving unit for deriving a defect candidate determination value to be
A difference value calculation unit for calculating a difference value between the luminance value of the target pixel and the average luminance value;
And a defect candidate determination unit that determines whether or not the target pixel is a defect candidate pixel based on the difference value and the defect candidate determination value.   In the setting condition information, the predetermined range is a first position that is farther from the target pixel than the pixel adjacent to the target pixel, and a second position that is farther from the target pixel than the first position. The defect determination apparatus according to claim 1, wherein the information is a range up to a range up to. A correspondence information creating device that creates the correspondence information in the defect determination device according to claim 1 based on a model product selected in advance as a model of the product to be inspected.
A model-target pixel selection unit that selects one of the plurality of pixels in an image including a plurality of pixels representing the model product as a target pixel;
A reference pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the target pixel, the pixel group not including the target pixel and a pixel adjacent to the target pixel A model reference pixel group extraction unit for extracting a pixel group;
An exemplary product average luminance value calculating unit for calculating an average luminance value of all the pixels constituting the reference pixel group;
An exemplary product difference value calculation unit for calculating a difference value between the luminance value of the target pixel and the average luminance value;
Based on the relationship between the average luminance value corresponding to each pixel of interest and the difference value, a defect candidate determination value corresponding to the average luminance value is calculated, and the correspondence relationship between the defect candidate determination value and the average luminance value A correspondence relationship information creating unit for creating correspondence relationship information indicating the correspondence relationship information. A defect determination method executed by an information processing apparatus,
A pixel-of-interest selection step in which the information processing apparatus selects one of the plurality of pixels in an image composed of a plurality of pixels representing an inspection target product as a pixel of interest;
The information processing apparatus is a pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the pixel of interest, and includes the pixel of interest and a pixel adjacent to the pixel of interest A reference pixel group extraction step for extracting a reference pixel group that is not a pixel group;
The information processing apparatus calculates an average luminance value of all the pixels constituting the reference pixel group, and calculates based on correspondence information indicating a correspondence relationship between a defect candidate determination value and an average luminance value, which is determined in advance. A defect candidate determination value derivation step for deriving a defect candidate determination value corresponding to the average brightness value,
A difference value calculating step in which the information processing apparatus calculates a difference value between a luminance value of the target pixel and the average luminance value;
A defect determination method comprising: a defect candidate determination step for determining whether or not the target pixel is a defect candidate pixel based on the difference value and the defect candidate determination value. . A correspondence information creation method executed by an information processing apparatus, which creates the correspondence information in the defect determination method according to claim 4 based on a model product selected in advance as a model of the product to be inspected. ,
The information processing device selects the one of the plurality of pixels in the image composed of a plurality of pixels representing the model product as a target pixel;
The information processing apparatus is a pixel group including a plurality of pixels within a predetermined range based on predetermined setting condition information around the pixel of interest, and includes the pixel of interest and a pixel adjacent to the pixel of interest A model reference pixel group extraction step for extracting a reference pixel group that is not a pixel group;
An exemplary product average luminance value calculating step in which the information processing apparatus calculates an average luminance value of all the pixels constituting the reference pixel group;
The information processing apparatus calculates a difference value between the luminance value of the target pixel and the average luminance value;
The information processing apparatus calculates a defect candidate determination value corresponding to the average luminance value based on a relationship between the average luminance value corresponding to each pixel of interest and the difference value, and the defect candidate determination value and the average A correspondence relationship information creating method, comprising: a correspondence relationship information creation unit for creating correspondence relationship information indicating a correspondence relationship with a luminance value.

Images