JP2008175558A - Device and method for visual inspection - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a device and method for visual inspection capable of stably, automatically, and efficiently detecting unevenness even when the unevenness with different sizes and shapes are mixed. <P>SOLUTION: A central area mean value calculating part 14 calculates the mean value (the central area mean luminance)of the luminance (the gradation values of each picture element in the imaging data) in the central area corresponding to the object picture elements. A peripheral area 1 mean value calculating part 151-a peripheral area N mean value calculating part 15N calculate the mean values of the grading values (a peripheral area N mean luminance) of each picture element at the peripheral area 1- the peripheral area N corresponding to the central area in the same way. Next, a mean difference calculating part 16 calculates the difference values (the difference value 1- the difference value N) between the mean luminance of the central area and that of the peripheral area 1-peripheral area N. Furthermore, the mean difference value calculating part 16 extracts the representative value (the representative difference value) among the difference values of the difference value 1- the difference value N. In this visual inspection device 1, detection of unevenness is achieved using this representative difference value. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an appearance inspection apparatus and an appearance inspection method, and more particularly to an appearance inspection apparatus and an appearance inspection method for inspecting a screen (appearance) of a display device and detecting unevenness.

  In manufacturing display devices, especially large-screen displays with a large number of pixels, technology for detecting and inspecting pixels for defects and irregularities on the display screen is important, and this inspection is automatically performed using a computer. It is essential for its production. In this inspection, an inspection image obtained by imaging the screen displayed on these display devices is obtained, and the computer analyzes defects and unevenness in the inspection image by analyzing the luminance (shading value) of each pixel in the inspection image. Recognize and detect patterns. However, regarding unevenness in particular, since the shape and strength (shading) vary, it is difficult to automatically identify this. For this reason, unless this pattern recognition is made appropriate, the situation where the nonuniformity to be detected is not detected or the region that is not the original nonuniformity is recognized as nonuniformity occurs, and an appropriate inspection result cannot be obtained.

  Further, for example, when the illumination at the time of imaging in the inspection image is not uniform or electrical noise exists, the inspection image is also affected by these. In the former case, low-frequency noise is superimposed on the inspection image, and in the latter case, high-frequency noise is superimposed. When low-frequency noise is superposed on the low-frequency noise, the gray value of each pixel changes gradually from one end of the image to the other. When high-frequency noise is superimposed, the gray value may fluctuate greatly between adjacent pixels due to this influence. Therefore, it is difficult to detect unevenness simply by examining the gray value of each pixel.

  Patent Document 1 describes an appearance inspection method that improves such a situation. Here, in the obtained inspection image, as shown in FIG. 15, a first area pattern including a certain pixel (target pixel) and a second area pattern around it are set, and the average density in each area is set. The value is calculated and the difference is calculated. This operation is performed with all pixels of the display device as target pixels, and emphasized image data in which this difference is mapped is created. If the target pixel is included in the uneven region, the difference becomes large, and if not included, the difference becomes small. For this reason, in this emphasized image data, a region having a large difference is recognized as uneven. That is, it is determined by this difference whether or not the target pixel is included in the uneven region. Here, the first region pattern and the second region pattern can be arbitrarily set, but by optimizing this shape according to the shape and size of the unevenness, unevenness is emphasized in this enhanced image data. The recognition becomes easy. Also, instead of simply using the gray value of each pixel, the average gray value in the first area pattern and the second area pattern is used, and further, by using these differences, low frequency noise and high frequency noise are obtained. The effect of is removed. Therefore, it was possible to automatically and efficiently perform unevenness detection corresponding to unevenness of various shapes and sizes.

JP 2004-69673 A

  However, unevenness that occurs on the same screen of the same display device is not always the same size and shape, and unevenness of different sizes and shapes often occur in a mixed manner. In such a case, when the above-described conventional image inspection method is used, it is necessary to perform processing for each type of unevenness. That is, it is necessary to classify the unevenness into a plurality of types for each size and shape, and to set the first region pattern and the second region pattern for each type to create emphasized image data. For this reason, when unevenness of different sizes and shapes coexists, the setting of the first region pattern and the second region pattern is complicated for the operator, and the processing time becomes longer in proportion to the number of types of unevenness. There was a problem of becoming.

  The present invention has been made in view of the above problems, and an object thereof is to provide an appearance inspection apparatus and an appearance inspection method that solve the above-described problems.

In order to solve the above problems, the present invention has the following configurations.
The gist of the invention described in claim 1 is an appearance inspection apparatus for detecting a nonuniform region in image data, a region setting means for setting a central region and a plurality of peripheral regions around the central region, and the image Center area average luminance calculating means for identifying a pixel in the data and calculating a center area average luminance that is an average value of gray values of each pixel in the center area corresponding to the pixel in the image data; and a center corresponding to the pixel A peripheral area average luminance calculating means for calculating a peripheral area average luminance that is an average value of gray values of each pixel in the peripheral area corresponding to the area for each of the plurality of peripheral areas, the central area average luminance, and the peripheral areas An appearance inspection apparatus comprising: an average difference calculating unit that calculates a difference value with respect to an average luminance for each of the peripheral regions and extracts a representative difference value from each of the difference values. Exist for.
The gist of the invention described in claim 2 is to determine whether or not the pixel is included in the uneven region based on the representative difference value in each pixel in the image data. It exists in the appearance inspection device.
The gist of the invention described in claim 3 is characterized in that the average difference calculating means extracts a maximum value among a plurality of difference values corresponding to the plurality of peripheral regions as a representative difference value. 2. It exists in the external appearance inspection apparatus of 2.
The gist of the invention described in claim 4 is that the priorities are set in the plurality of peripheral areas, and the average luminance in the peripheral areas is calculated according to the priorities. Exists in the device.
The gist of the invention described in claim 5 is that a difference threshold value in the difference value is set, the difference values corresponding to the plurality of peripheral regions are sequentially calculated in the specified pixel according to the priority order, and the difference value is When the difference threshold is exceeded, the difference value is extracted as the representative difference value, and the difference value corresponding to the peripheral region having a priority order later than the priority order of the peripheral region corresponding to the difference value is calculated. 5. The appearance inspection apparatus according to claim 4, wherein the inspection is not performed.
The gist of the invention described in claim 6 resides in the appearance inspection apparatus according to claim 4 or 5, wherein the priorities are set in the order of decreasing number of pixels occupied by the respective peripheral regions.
The gist of the invention described in claim 7 is that a representative difference threshold value in the representative difference value is set, and it is determined whether or not each pixel is included in the uneven region by comparing the representative difference threshold value with the representative difference value. 7. The appearance inspection apparatus according to claim 1, further comprising an uneven area extraction unit.
The gist of the invention described in claim 8 resides in the appearance inspection apparatus according to claim 7, wherein an assembly of pixels determined to be included in the uneven region is determined to be uneven.
The gist of the invention described in claim 9 is the appearance inspection apparatus according to claim 8, further comprising unevenness measuring means for measuring the size of the unevenness.
The gist of the invention described in claim 10 resides in the appearance inspection apparatus according to claim 9, wherein the unevenness measuring means measures the number of pixels constituting the unevenness.
The gist of the invention described in claim 11 resides in the appearance inspection apparatus according to any one of claims 1 to 10, wherein the plurality of peripheral regions are annular and do not overlap each other.
The gist of the invention described in claim 12 is an appearance inspection method capable of detecting unevenness in image data, wherein a central region and a plurality of peripheral regions around the central region are set, and the image data A pixel is specified, a central area average luminance that is an average of the gray values of each pixel in the central area corresponding to the pixel is calculated, and the gray value of each pixel in the peripheral area corresponding to the central area corresponding to the pixel is calculated. An average peripheral area average luminance is calculated for each of the plurality of peripheral areas, a difference value between the central area average luminance and each of the peripheral area average luminances is calculated for each of the peripheral areas. The visual inspection method is characterized in that representative difference values are extracted from.
The gist of the invention described in claim 13 resides in the appearance inspection method according to claim 12, wherein unevenness is detected based on the representative difference value calculated for each pixel in the image data.
The gist of the invention described in claim 14 is that the maximum value among a plurality of difference values corresponding to the plurality of peripheral regions is extracted as a representative difference value. Exist.
The gist of the invention described in claim 15 is that the priority order is set in the plurality of peripheral areas, and the peripheral area average luminance is calculated according to the priority order. Exist.
The gist of the invention of claim 16 is a case where a difference threshold value in the difference value is set, the difference value corresponding to the plurality of peripheral regions is calculated according to the priority, and the difference value exceeds the difference threshold value. Is characterized in that the difference value is extracted as the representative difference value, and the difference value corresponding to a peripheral region having a priority after the priority of the peripheral region corresponding to the difference value is not calculated. The present invention resides in an appearance inspection method according to claim 15.
The gist of the invention described in claim 17 resides in the appearance inspection method according to claim 15 or 16, characterized in that the priority is set in the order of decreasing number of pixels occupied by the respective peripheral regions.
The gist of the invention described in claim 18 is that a representative difference threshold value in the representative difference value is set, and it is determined whether or not each pixel is included in the uneven region by comparing the representative difference threshold value with the representative difference value. It exists in the external appearance inspection method of any one of Claim 12 thru | or 17 characterized by the above-mentioned.
The gist of the invention described in claim 19 resides in the appearance inspection method according to claim 18, characterized in that an assembly of pixels determined to be included in the uneven region is determined to be uneven.
The gist of the invention described in claim 20 resides in the appearance inspection method according to claim 19, characterized in that the size of the unevenness is measured.
The gist of the invention described in claim 21 resides in the appearance inspection method according to claim 19 or 20, wherein the number of pixels constituting the unevenness is measured.
The gist of the invention described in claim 22 resides in the appearance inspection method according to any one of claims 12 to 21, wherein the plurality of peripheral regions are annular and do not overlap each other.

  ADVANTAGE OF THE INVENTION According to this invention, even when the nonuniformity from which a magnitude | size and a shape coexist, the external appearance inspection apparatus and external appearance inspection method which can detect an unevenness stably and automatically can be provided.

(First embodiment)
Hereinafter, an appearance inspection apparatus and an inspection method according to a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a configuration diagram of an appearance inspection apparatus according to the present embodiment. The appearance inspection apparatus 1 is input with two-dimensional image data to be inspected, that is, data indicating the luminance (shading value) of each pixel in the two-dimensional image. This image data is obtained by taking an image of a display screen of a display device (not shown) to be inspected, such as a plasma display or a liquid crystal display, and is data of an inspection image in which unevenness is to be detected. is there. This imaging method is not particularly limited as long as it can detect a change in luminance between pixels in the unevenness to be detected. For example, if a camera using a solid-state imaging device is used, a high-resolution and high-sensitivity image can be obtained. Therefore, it is particularly preferable. Depending on the type of unevenness, this image may be a monochrome image or a color image. In the latter case, for example, the following processing may be performed for each type as three types of image data for each RGB, or the subsequent processing may be performed using the converted amount of the signal for each RGB. .

  This appearance inspection apparatus 1 detects unevenness. The brightness of the inside of this uneven area in the inspection image is higher than that of the outside. For this reason, the nonuniformity area detected here is an area in the image data that is recognized as having a higher luminance than the surrounding area.

  FIG. 2 is a flowchart showing this appearance inspection method. Hereinafter, the appearance inspection method will be described with reference to FIGS. 1 and 2.

  First, in the appearance inspection apparatus 1 shown in FIG. 1, the image data is input to the image input unit 11 (S1).

  On the other hand, the region setting unit 12 is connected to the input unit 13. The input unit 13 includes a keyboard, a display, and the like, and a user (inspector) can arbitrarily set the shape and size of the central region used for unevenness detection in the image data using these. (S2). This setting is made in units of pixels in the image data.

  Similarly, the user can arbitrarily set the shape and size of N types of peripheral regions (peripheral region 1 to peripheral region N), and inputs them (S3). This setting is also made in units of pixels. This number N can be arbitrary.

  FIG. 3 shows an example of the relationship between the central region and the peripheral region. In this example, one cell corresponds to a pixel in the image data, and there are three types of peripheral regions (N = 3). Here, the gray area of A is the central area, the gray area of B is the peripheral area 1, the gray area of C is the peripheral area 2, and the gray area of D is the peripheral area 3. In addition, a target pixel described later is a central pixel (black) in the central region. As in this example, each peripheral region has an annular shape surrounding a central region (3 × 3 pixels), but it is preferable that the peripheral regions do not overlap each other. Note that the shape of the central region, the positional relationship between the target pixel and the central region, and the positional relationship between the central region and each peripheral region can be arbitrarily set. For example, in this example, the central region has a 3 × 3 pixel configuration including the target pixel. However, the present invention is not limited to this. For example, the central region may be annular and the target pixel may be set as the center. . In this case, the target pixel is not included in the central region.

  The central area setting (S2) and the peripheral area setting (S3) may be performed before the image data input (S1).

  Next, the center area average calculation unit 14 sets one pixel (target pixel) in the image data (S4), and performs the following calculation. That is, the following calculation is calculated individually for each pixel in the image data, and processing is performed for all the pixels.

  The center area average calculation unit 14 calculates the average value (center area average brightness) of the brightness (the gray value of each pixel in the image data) in the center area corresponding to the target pixel (S5). For example, when the center region shown in FIG. 3 is set, the average value of the gray values of the pixels in 3 × 3 pixels centering on the target pixel is calculated.

  Similarly, the peripheral region 1 average calculation unit 151 to the peripheral region N average calculation unit 15N similarly calculates the average value (peripheral region N average luminance) of the gray values of the pixels in the peripheral region 1 to the peripheral region N corresponding to the central region. Calculate (S6).

  Next, the average difference calculation unit 16 calculates a difference value (difference value 1 to difference value N) between the central region average luminance and the peripheral region 1 average luminance to the peripheral region N average luminance. That is, (difference value 1 = central region average luminance−peripheral region 1 average luminance) to (difference value N = central region average luminance−peripheral region N average luminance) is calculated (S7).

  Furthermore, the average difference calculation unit 16 extracts a representative value (representative difference value) from the difference values 1 to N. Here, the representative difference value is the maximum value of (difference value 1, difference value 2,..., Difference value N). This representative difference value is stored in the memory (S8). In this appearance inspection apparatus 1, unevenness is detected using this representative difference value.

  A threshold value (representative difference threshold value) for the representative difference value is set in advance in the unevenness region extraction unit 17. The unevenness area extraction unit 17 determines whether or not the representative difference value is larger than the representative difference threshold value. The pixel determined to be large is determined to be included in the uneven region, and the determination result is stored in the memory (S9). Here, the data to be stored are the coordinates (position) of the target pixel in the image data and the determination result.

  Next, it is determined whether or not the processing of S5 to S9 has been performed for all the pixels in the image data (S10). If the process has not been performed for all the pixels, the next pixel is set (S4), and the processes after S5 are performed again.

  In the present embodiment, unevenness is easily detected in an image in which representative difference values in each pixel are mapped. This principle will be described below.

  The brightness inside the uneven area is higher than the brightness outside. That is, when the target pixel is included in the uneven region, the brightness of the pixels near the target pixel is higher than the brightness of the surrounding pixels. Therefore, it is possible to determine whether or not the target pixel is included in the uneven region by examining the difference between the brightness of the pixels near the target pixel and the brightness of the surrounding pixels. However, which range is defined as the vicinity of the target pixel and which range is defined as the periphery thereof cannot be uniquely determined, and the optimum range for this detection varies depending on the shape and size of the unevenness.

  In the present embodiment, a central region is defined as a range in the vicinity of the target pixel. Here, as shown in FIG. 3, it is assumed that the target pixel is the center of the central region. On the other hand, N types of regions of the peripheral region 1 to the peripheral region N are defined as the periphery thereof. For this reason, the average brightness in these areas is calculated, and the difference in average brightness between the central area and the peripheral areas 1 to N is calculated.

  An example in which this inspection method is applied to actual unevenness is shown in FIGS. Here, it is assumed that the setting of each area is as shown in FIG. 3 (N = 3), and the uneven area to be detected is indicated by a rectangular shape indicated by a broken line in each figure. 4 to 6, the size of the unevenness is different, and is small (FIG. 4), medium (FIG. 5), and large (FIG. 6). In addition, the shading value of each pixel is 1 inside the uneven region, and is simplified to 0 outside. In any of FIGS. 4 to 6, the position of the target pixel (black) is illustrated with respect to three types of X (inside the unevenness), Y (near the uneven boundary), and Z (outside the unevenness). In this case, if (value at X)> (value at Y)> (value at Z) in a certain amount, this amount is used to identify whether the target pixel is included in the uneven region. Is possible.

First, when the unevenness is small (FIG. 4), the table in the figure shows the central area average luminance (I _A ), the peripheral area 1 average luminance (I _B ), the peripheral area 2 average luminance (I _C ), and the peripheral area 3. Average brightness (I _D ) is shown for each position (X, Y, Z). The bottom of the table were calculated by this difference value ₁ (I A _-I B), the difference value ₂ (I A _-I C), the difference value _{3 (I} A -I _D) are shown. In any of the difference values 1 to 3, (value in X)> (value in Y)> (value in Z). Therefore, the uneven region can be identified using any one of the difference values 1 to 3.

  Next, the results of the same calculation when the unevenness is medium (FIG. 5) are shown in the table of FIG. In the case of the difference value 2 and the difference value 3, a value that satisfies the condition for identifying the uneven region is output, but since the difference value 1 indicates a small value (0) at the position X, this condition is satisfied. Absent. That is, the difference value 1 cannot be used as a value for identifying the unevenness of this size. This indicates that the size of the peripheral region 1 (B) is inappropriate because it is too small as a peripheral region used for detecting the unevenness of this size.

  Next, the results of similar calculations when the unevenness is large (FIG. 6) are shown in the table of FIG. In the case of the difference value 3, a value that satisfies the condition for identifying the uneven area is output, but in the case of the difference value 1 and the difference value 2, this condition is not satisfied. That is, the difference value 1 and the difference value 2 cannot be used as values for identifying the unevenness in size. This indicates that the peripheral region 1 and the peripheral region 2 are not suitable as the peripheral region used for detecting the unevenness of this size.

  On the other hand, in FIGS. 4 to 6, representative difference values (that is, the maximum value of the difference values 1 to 3) are shown at the bottom of the table. In particular, in the case of FIG. 5 and FIG. 6, even if there is a difference value 1 to difference value 3 that is not suitable for unevenness identification, in the case where the representative difference value is used, both of these are identified. Is possible. The peripheral area from which the representative difference value is obtained is suitable for detection of unevenness of each size, that is, selecting the maximum value from the difference value 1 to the difference value N This corresponds to the selection of the one suitable for unevenness detection from the peripheral area N. Therefore, in this embodiment, if the center region and the peripheral region 1 to the peripheral region N are set first without assuming the size of the unevenness in advance (S2, S3), unevenness of various sizes is mixed. Even in this case, a result equal to that the optimum one is automatically selected from the peripheral regions 1 to N according to the respective sizes can be obtained.

  Further, as in this example, it is preferable that each peripheral region has an annular shape surrounding the central region, and the sizes thereof are different. However, the shape is arbitrary, and although it is a rectangular ring in FIG. 3, it may be, for example, a circular ring. Further, it is preferable that the number of pixels constituting each peripheral region is small because the calculation time can be shortened. For this reason, in the example of FIG. 3, each peripheral region is configured with a minimum line width (unit pixel).

  In the above example, the case where the unevenness is all rectangular and the sizes thereof are different has been described. However, it is obvious that the same effect can be obtained even when unevenness having different shapes are mixed. Therefore, when the representative difference value obtained by the above calculation is mapped, unevenness can be easily identified regardless of its shape and size.

  When the processing of S5 to S9 is performed for all the pixels, a nonuniform image in which the representative difference values are mapped is displayed (S11). As a nonuniformity image, for example, a pixel whose representative difference value is larger than the representative difference threshold value is black, and a pixel whose representative difference threshold value is less than or equal to the representative difference threshold value is two-dimensionally displayed on the display. In the examples of FIGS. 4 to 6, for example, if the representative difference threshold is set to a value of about 0.3, unevenness of any size can be detected. Even when there are many unevennesses having different shapes and sizes in the image data, each of these unevenness images is displayed with being emphasized.

  The unevenness measurement unit 18 determines that the aggregate is uneven when a plurality of pixels determined to be included in the unevenness region are adjacent to each other in the unevenness image, and measures the feature amount of each unevenness (S12). ). As the unevenness feature amount, an amount indicating the unevenness size and the uneven shade value is measured. As the size of the unevenness, for example, the major axis, the minor axis, the peripheral length, the number of pixels constituting the unevenness, and the like are measured. As the uneven density value, the maximum value of the representative difference values in the pixels constituting the unevenness can be used, and the average value of the representative difference values in the pixels constituting the unevenness can also be used. The unevenness measurement unit 18 outputs these feature amounts to the unevenness determination unit 19.

  The non-uniformity determination unit 19 compares the measured non-uniformity feature amount with a predetermined criterion for determining the non-uniformity feature amount (S13), and outputs the determination result (S14). For example, if the uneven perimeter of the image data to be evaluated is greater than or equal to the set value, or if the uneven gray value is greater than or equal to the set value, the image that is the source of this image data is displayed. The displayed display is determined to be defective.

  As described above, this visual inspection apparatus or visual inspection method can automatically and efficiently detect unevenness, and can easily identify whether or not the display device to be inspected is a defective product.

  An example in which this appearance inspection apparatus or appearance inspection method is applied to an inspection image on which low-frequency noise is superimposed will be described. Here, as shown in the left of FIG. 7, the inspection image has a tendency to be dark on the left side and bright on the right side, and includes unevenness to be detected. Here, the profile of the luminance in the horizontal direction (shading value) in the vertical coordinate value L including the uneven region is shown in FIG. Here, the luminance gradually changes from dark to bright from left to right over the entire line, and there is an uneven portion with a width wd pixel in the line.

  In the case where an uneven portion is simply extracted based on the luminance of each pixel, the threshold value is identified as d, for example. In this case, as shown in FIG. 7 (2), in addition to the uneven portion, the right portion of the line whose luminance is larger than the threshold value d is also recognized as uneven, which is not appropriate.

  On the other hand, when this inspection apparatus and inspection method are used, as shown in FIG. 8, in any part of D (outside unevenness), E (region including unevenness), and F (outside unevenness). The representative difference value, which is one of the differences between the central area average luminance and each peripheral area average luminance, is used, and the influence of the low frequency noise is cancelled. Accordingly, when this representative difference value is displayed in the same manner, as shown in FIG. 9, only the unevenness is emphasized, and the influence of low-frequency noise is removed.

  Next, an example in which this appearance inspection apparatus or appearance inspection method is applied to an inspection image on which high-frequency noise is superimposed will be described. Here, as shown in the left side of FIG. 10, the inspection image has its luminance (shading value) greatly fluctuated in each pixel due to noise, and this includes unevenness to be detected. Here, the luminance profile in the horizontal direction in the vertical coordinate value L including the uneven region is shown in FIG. Here, the luminance varies finely over the entire line, and an uneven portion with a width wd pixel exists therein.

  Also in this case, when the unevenness is identified using the threshold value in the luminance of each pixel, as shown in FIG. 10 (2), the original nonuniformity region is recognized as unevenness as in the case of the low-frequency noise. It is clear that there are things that are sometimes done.

  On the other hand, when this inspection apparatus and inspection method are used, as shown in FIG. 11, the central region average luminance is obtained at any point of G (outside of unevenness) and H (region including unevenness). A representative difference value that is one of the differences from the average luminance of each peripheral area is used. Here, since the average luminance in each region is used, the influence of the high frequency noise is removed by averaging. For this reason, when this representative difference value is displayed in the same manner, as shown in FIG. 12, only the unevenness is emphasized, and the influence of high-frequency noise is removed.

  Therefore, according to the above-described appearance inspection apparatus and appearance inspection method, even when a plurality of types of unevenness having different shapes and sizes are mixed, it is easy to detect each type of unevenness.

  In the examples of FIGS. 4 to 6, N = 3, but the number can be arbitrary. Increasing N can be widely optimized for various types of unevenness, and the unevenness detection capability becomes higher. . On the other hand, when N is smaller, the calculation time can be shortened and the memory of the computer to be used can be saved. Therefore, if the range of the shape and size of the unevenness detected in advance is known, the number can be minimized by setting appropriate peripheral regions 1 to N corresponding to this. ,preferable.

(Second Embodiment)
Next, an appearance inspection apparatus and inspection method according to a second embodiment of the present invention will be described with reference to the drawings. FIG. 13 is a configuration diagram of an appearance inspection apparatus 30 according to the present embodiment. FIG. 14 is a flowchart showing this inspection method. Hereinafter, the appearance inspection apparatus and the appearance inspection method will be described based on these.

  First, in the appearance inspection apparatus 30, the image data is input to the image input unit 31 (S21).

  On the other hand, the region setting unit 32 is connected to the input unit 33, and the user (inspector) uses these to set a central region used for unevenness detection in the image data (S22).

  Similarly, the user can arbitrarily set the shape and size of N types of peripheral areas (peripheral areas 1 to N), and inputs them (S23).

  The above process and each component are not different from those of the first embodiment, and detailed description thereof is omitted. In the appearance inspection apparatus and the appearance inspection method according to the second embodiment, the processes and functions of constituent elements described below are particularly different.

  The peripheral area selection unit 36 determines the order in which the set peripheral areas 1 to N are applied in the subsequent calculations, that is, the priority order (S24). Here, the reference for determining the priority order is the number of pixels constituting these. That is, a higher priority is set in order from the smallest number of pixels. For example, in the example of FIG. 3, the peripheral region 1 (B) having 24 pixels is ranked first, the peripheral region 2 (C) having 40 pixels is ranked second, and the peripheral region 3 having 56 pixels (D) is third.

  Next, the center area average calculation unit 34 sets one pixel (target pixel) in the image data (S25), and performs the following calculation. That is, the following calculation is calculated individually for each pixel in the image data, and is calculated for all pixels.

  The center area average calculation unit 34 calculates the average value (center area average brightness value) of the brightness (pixel gray value of the image data) in the center area corresponding to the target pixel (S26). Steps S25 and S26 are the same as S4 and S5 in FIG. 2 in the first embodiment.

  Next, the peripheral region selection unit 36 selects and sets the peripheral region (applied peripheral region) used in the subsequent calculations from the peripheral region 1 to the peripheral region N (S27). This setting is performed according to the above-mentioned priority order. In other words, the peripheral area having the first priority is the application peripheral area.

  Next, the peripheral area average calculating unit 35 calculates the average value (applied peripheral area average luminance) of each pixel in the applied peripheral area corresponding to the central area (S28).

  Next, the average difference determination unit 37 calculates a difference value between the central region average luminance value and the applied peripheral region average luminance value (S29). Here, a difference threshold value corresponding to this difference value is set in advance, and it is determined whether or not this difference value is equal to or greater than this difference threshold value (S30). If the difference is equal to or greater than the difference threshold, the determination result is stored in the memory, and the process for the target pixel is completed. Therefore, it is determined whether or not the processing of S26 to S29 has been performed for all the pixels in the image data (S32). If the process has not been performed for all the pixels, the next pixel is set (S25), and the processes after S26 are performed again.

  Here, the difference threshold value is set and stored in advance from the input unit 33 or the like, and the average difference determination unit 37 reads this and performs the above-described processing. That the difference value is greater than or equal to the difference threshold corresponds to the determination that the target pixel is included in the uneven region. For this reason, when this determination is made, calculation using another peripheral region is not necessary, and the processing for this target pixel is completed.

  If this difference value is smaller than the difference threshold value (S30), it is determined whether or not the processing of S27 to S30 has been performed for all the peripheral regions (peripheral region 1 to peripheral region N) (S31). If the processing has not been performed for all the peripheral regions, the next applicable peripheral region is set (S27), and the subsequent processing is performed. That is, a new application peripheral area is set for this target pixel, and a new calculation is performed. Here, the next applicable peripheral area is a peripheral area having the next priority.

  Therefore, in this embodiment, it is not necessary to perform calculation for all of the set peripheral areas 1 to N, and when it is determined that the peripheral areas are sequentially selected and the target pixel is included unevenly, The process for the target pixel ends. If the difference value does not exceed the difference threshold even when all of the peripheral regions 1 to N are applied, it is determined that the target pixel is not included unevenly, and the determination result Is stored in the memory.

  Therefore, after the processing for all the pixels is completed (S32), the determination result as to whether or not the pixels are included unevenly is stored in the memory for all the pixels. Therefore, it is possible to create a two-dimensional image (uneven image) in which the determination result is displayed in black and white, for example (S33).

  Note that, as in the first embodiment, it is possible to perform measurement of uneven feature values, pass / fail determination, and the like based on this uneven image. However, since these are the same as those in the first embodiment, here I omitted it.

  The case where the appearance inspection apparatus and the appearance inspection method according to the second embodiment are applied to the cases of FIGS. In this case, the priority order in each peripheral region is the first in the peripheral region 1 (B), the second in the peripheral region 2 (C), and the third in the peripheral region 3 (D). The difference threshold value can be set to 0.3, for example, similarly to the representative difference threshold value.

  When the unevenness is small (FIG. 4), since the difference value 1 calculated at the first time already exceeds the difference threshold at the position X, the peripheral region 2 (C) and the peripheral region 3 (D) Is not calculated. At the position Y, since this difference threshold is exceeded at the second time, the calculation for the peripheral region (D) is not performed. For the position Z, since no difference value exceeds the difference threshold value, calculation is performed for all the surrounding areas. Therefore, three of the three types of difference values (total of nine) at the three positions in the table of FIG. 4 are not calculated.

  Similarly, when the size of the unevenness is medium (FIG. 5), the calculation for the peripheral region 3 (D) at the positions X and Y is not performed. Accordingly, two of the three types of difference values (total of nine) at the three positions in the table of FIG. 5 are not calculated.

  When the unevenness is large (FIG. 6), it is the peripheral region 3 (D) at the position X and the peripheral region 3 at the position Y that exceed the difference threshold value, so that 3 at three positions in the table of FIG. All kinds of difference values (total of 9) are calculated.

  Therefore, in 27 examples of calculating the difference value 1 to the difference value 3 in the examples of FIGS. For this reason, there is an effect that the processing time is shortened as compared with the case of the first embodiment. In this case, the number N of peripheral regions is 3, but it is obvious that this effect is further increased when N is large.

  Even with the appearance inspection apparatus 30 and the inspection method, as in the first embodiment, even when a plurality of types of unevenness having different shapes and sizes are mixed, it is easy to detect each type of unevenness. is there. Furthermore, since it is not necessary to calculate the average luminance value for all of the set peripheral areas 1 to N, there is an effect that the total processing time is shortened. The peripheral area with a smaller number of pixels requires a shorter time for calculating the average luminance value, and the peripheral area with a smaller number of pixels is preferentially selected. Therefore, in the second embodiment, this processing time is particularly shortened. Can do. This effect is particularly effective when the number N of peripheral regions is large.

  In the above example, the priority order for selecting the peripheral region is set in the order of decreasing the number of pixels. However, this priority order can be appropriately set according to the situation. For example, when the most frequently detected type of unevenness can be predicted in advance, the same effect can be obtained by increasing the priority of the peripheral region most suitable for detecting this unevenness.

  In the first and second embodiments, the above-described processing is performed for all the pixels in the image data. If an area in which unevenness is assumed to be known is known in advance, this area is used. Unevenness can be detected more efficiently by performing the above-described processing only for.

  In the above example, one type of central region is used and N types of peripheral regions are used. However, it is also possible to perform the same calculation by setting the central region to a plurality of types having different sizes and / or shapes. is there. There can be a plurality of types of central regions and a single type of peripheral region.

  In the above example, the non-uniformity to be detected is an area whose brightness is higher than that of the peripheral part in the inspection image, but conversely, an area whose brightness is smaller than that of the peripheral area should be set to be non-uniform. You can also. Even in this case, it is obvious that the unevenness can be similarly detected by reversing the magnitude relationship of the difference value or the like in the above example.

It is a lineblock diagram of the appearance inspection device concerning a 1st embodiment of the present invention. It is a flowchart which shows the external appearance inspection method which concerns on the 1st Embodiment of this invention. It is an example of the center area | region and peripheral area | region set in the external appearance inspection method which concerns on the 1st Embodiment of this invention. This is an example in which the appearance inspection method according to the first embodiment of the present invention is applied when the unevenness is small. This is an example in which the appearance inspection method according to the first embodiment of the present invention is applied when the unevenness is medium. This is an example in which the appearance inspection method according to the first embodiment of the present invention is applied when the unevenness is large. It is an example of the test | inspection image when a low frequency noise is superimposed. 8 is an example in which the appearance inspection method according to the first embodiment is applied to the inspection image of FIG. 7. It is the example which calculated the representative difference value in the test | inspection image of FIG. It is an example of the test | inspection image when a high frequency noise is superimposed. It is an example which applied the appearance inspection method which concerns on 1st Embodiment to the test | inspection image of FIG. It is the example which calculated the representative difference value in the test | inspection image of FIG. It is a block diagram of the external appearance inspection apparatus which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows the external appearance inspection method which concerns on the 2nd Embodiment of this invention. It is a figure which shows an example of the 1st area | region pattern and 2nd area | region pattern which are set in the conventional external appearance inspection method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 30 Appearance inspection apparatus 11, 31 Image input part 12, 32 Area setting part 13, 33 Input part 14, 34 Central area average calculation part 151-15N Peripheral area 1 average calculation part-Peripheral area N average calculation part 16 Average difference Calculation unit 17, 38 Uneven region extraction unit 18, 39 Uneven measurement unit 19, 40 Unevenness determination unit 35 Peripheral region average calculation unit 36 Peripheral region selection unit 37 Average difference determination unit

Claims (22)

An appearance inspection apparatus for detecting uneven areas in image data,
A region setting means for setting a central region and a plurality of peripheral regions around the central region;
A central area average luminance calculating unit that identifies a pixel in the image data and calculates a central area average luminance that is an average value of gray values of pixels in the central area corresponding to the pixel in the image data;
A peripheral area average luminance calculating means for calculating a peripheral area average luminance that is an average value of gray values of each pixel in the peripheral area corresponding to the central area corresponding to the pixel for each of the plurality of peripheral areas;
Calculating a difference value between the central region average luminance and each peripheral region average luminance for each peripheral region, and extracting a representative difference value from the difference values;
A visual inspection apparatus comprising:   The appearance inspection apparatus according to claim 1, wherein it is determined whether or not the pixel is included in a nonuniform region based on the representative difference value of each pixel in the image data.   The appearance inspection apparatus according to claim 1, wherein the average difference calculation unit extracts a maximum value among a plurality of difference values corresponding to the plurality of peripheral regions as a representative difference value.   The appearance inspection apparatus according to claim 1, wherein a priority order is set in the plurality of peripheral areas, and the peripheral area average luminance is calculated according to the priority order. A difference threshold in the difference value is set,
In the specified pixel, the difference values corresponding to the plurality of peripheral regions are sequentially calculated according to the priority order, and when the difference value exceeds the difference threshold, the difference value is extracted as the representative difference value. 5. The appearance inspection apparatus according to claim 4, wherein a difference value corresponding to a peripheral area having a priority after the priority of the peripheral area corresponding to the difference value is not calculated.   6. The appearance inspection apparatus according to claim 4, wherein the priority order is set in an order of decreasing number of pixels occupied by the peripheral regions. A representative difference threshold in the representative difference value is set,
7. The uneven region extracting means for determining whether or not each pixel is included in the uneven region by comparing the representative difference threshold value with the representative difference value. Item 1. An appearance inspection apparatus according to item 1.   The appearance inspection apparatus according to claim 7, wherein the collection of pixels determined to be included in the uneven region is determined to be uneven.   The appearance inspection apparatus according to claim 8, further comprising unevenness measuring means for measuring the size of the unevenness.   The visual inspection apparatus according to claim 9, wherein the unevenness measuring unit measures the number of pixels constituting the unevenness.   The appearance inspection apparatus according to claim 1, wherein the plurality of peripheral regions are annular and do not overlap each other. An appearance inspection method capable of detecting unevenness in image data,
Set a central area and a plurality of peripheral areas around the central area,
Identify a pixel in the image data, and calculate a central area average luminance that is an average of the gray value of each pixel in the central area corresponding to the pixel;
Calculating a peripheral area average luminance that is an average of the gray values of each pixel in the peripheral area corresponding to the central area corresponding to the pixel for each of the plurality of peripheral areas;
A difference value between the central area average luminance and each peripheral area average luminance is calculated for each peripheral area,
A visual inspection method, wherein a representative difference value is extracted from each of the difference values.   The visual inspection method according to claim 12, wherein unevenness is detected based on the representative difference value calculated for each pixel in the image data.   14. The appearance inspection method according to claim 12, wherein a maximum value among a plurality of difference values corresponding to the plurality of peripheral regions is extracted as a representative difference value.   14. The visual inspection method according to claim 12, wherein priority is set in the plurality of peripheral areas, and the average luminance of the peripheral area is calculated according to the priority. A difference threshold in the difference value is set,
The difference values corresponding to the plurality of surrounding areas are calculated according to the priority order, and when the difference value exceeds the difference threshold, the difference value is extracted as the representative difference value and corresponds to the difference value The visual inspection method according to claim 15, wherein a difference value corresponding to a peripheral area having a priority after the priority of the peripheral area to be calculated is not calculated.   The visual inspection method according to claim 15 or 16, wherein the priorities are set in an order of decreasing number of pixels occupied by the peripheral regions. A representative difference threshold in the representative difference value is set,
18. The appearance inspection method according to claim 12, wherein it is determined whether or not each pixel is included in a nonuniform region by comparing the representative difference threshold value with the representative difference value. .   The visual inspection method according to claim 18, wherein the collection of pixels determined to be included in the uneven region is determined to be uneven.   The appearance inspection method according to claim 19, wherein the size of the unevenness is measured.   21. The appearance inspection method according to claim 19, wherein the number of pixels constituting the unevenness is measured.   The appearance inspection method according to claim 12, wherein the plurality of peripheral regions are annular and do not overlap each other.