WO2010029932A1

WO2010029932A1 - Visual examination apparatus

Info

Publication number: WO2010029932A1
Application number: PCT/JP2009/065711
Authority: WO
Inventors: 昌年笹井
Original assignee: 株式会社メガトレード
Priority date: 2008-09-09
Filing date: 2009-09-09
Publication date: 2010-03-18
Also published as: KR20110040965A; JPWO2010029932A1; CN102138068A; CN102138068B; JP5084911B2

Abstract

In a case where a color image is used to examine the formation state of an object to be examined, the examination quality can be improved by establishing optimum values as examination threshold values for acquired RGB information of the color image. A visual examination apparatus (1) for examining an object (11) comprises an RGB information acquiring means (3) that acquires a color image from the object (11) to acquire RGB intensity data of examined portions of the object (11); a storing means (5) that stores therein RGB reference intensity data of the examined portions in an RGB polar coordinate system the axes of which indicate the respective intensity values; and a determining means (8) that compares the RGB intensity data of the examined portions acquired by the RGB information acquiring means (3) with the RGB reference intensity data of the RGB polar coordinate system stored in the storing means (5) to determine whether the RGB intensity data of the object (11) is included in the RGB reference intensity data, thereby determining the quality of the examined areas.

Description

Appearance inspection device

The present invention relates to an appearance inspection apparatus that acquires a color image from an inspection object and inspects the quality of the inspection object and an inspection method thereof, and more specifically, RGB luminance data of a color image is appropriately obtained. The present invention relates to an appearance inspection apparatus which can be inspected with high accuracy by setting a value.

Generally, the formation state of printed circuit boards, semiconductor wafers, liquid crystal substrates, etc. is inspected by an appearance inspection apparatus. When inspecting an inspection object by such an appearance inspection apparatus, generally, an image of the surface is acquired by a camera, and the quality of the inspection object is inspected from the acquired image. In particular, in recent years, when performing such inspection, a color image is acquired from an inspection object, and inspection is performed based on RGB information of the color image (Patent Documents 1 to 3, etc.). .

In such an appearance inspection apparatus, an example of inspecting a printed circuit board will be described. First, when inspecting a printed circuit board, images of pads, wiring patterns, resists, silk, etc. formed on the surface from the printed circuit board To get. Then, after aligning the image of the inspection object and the reference image, the pixel of the reference image corresponding to the pixel existing at a certain coordinate position of the inspection object is found, and within the allowable luminance width of the found reference pixel If the luminance value of the pixel of the inspection object is included in the pixel, it is determined that the pixel is a good pixel. Conversely, if it is not within the allowable luminance width, it is determined as a defective pixel, and if a predetermined number or more of defective pixels exist adjacently, it is determined that the printed circuit board is defective. (Patent Document 4).
JP 2007-101415 A JP 2006-78301 A JP 2006-78300 A JP 2007-309703 A

However, when such an inspection is performed with a color image, the following problems occur. That is, when a color image is acquired from an inspection object and RGB luminance data is inspected for each pixel, it is determined whether or not the pixel of the inspection object is included in each RGB reference luminance width. However, for example, when a pattern exists in the resist lower layer, even if it is a non-defective product, the RGB tint changes slightly due to uneven resist thickness, lot changes, and color mixing conditions. It may end up. In particular, when the luminance change is large, even if it is a non-defective product, the luminance width from dark green to bright green is in the range of (R, G, B) = (20, 120, 60) to (40, 200, 150). In the RGB Cartesian coordinate system, the threshold luminance width is set to a large value such as 20 <R (Δx) <40, 120 <G (Δy) <200, 60 <B (Δz) <150. (See FIG. 9). However, if the threshold luminance width is set to be large in this way, there is a possibility that a portion that should be determined as defective cannot be determined as defective. Specifically, for example, when there is a defective portion where the resist is peeled off and the copper of the pattern is exposed, generally, such a portion is covered with a thin resist, so that the complete copper color is not used. The resist color is slightly reddish (luminance values are (R, G, B) = (35, 150, 80)). For this reason, when the luminance width of the reference luminance data is increased, the rectangular parallelepiped area in FIG. 9 is increased, and the luminance value of the exposed portion is included in the luminance width of the reference luminance data and can be determined as defective. Disappear. On the other hand, if the threshold value is narrowed so that the exposed portion can be determined as defective, a non-defective product is determined as a defective product, and the subsequent visual inspection takes time.

Therefore, the present invention has been made paying attention to the above problems, and when inspecting the formation state of an inspection object using a color image, a threshold value for inspection is set for RGB of the acquired color image. An object of the present invention is to provide an appearance inspection apparatus which can be set to an optimum value and improve the quality of inspection.

In order to solve the above problems, the present invention obtains a color image from an inspection object in an appearance inspection apparatus that inspects the formation state of the inspection object on the basis of an image acquired from the inspection object. RGB information acquisition means for acquiring RGB luminance data in the inspection portion of the object, reference data storage means for storing RGB reference luminance data of each inspection portion in the RGB polar coordinate system with the axial direction as the luminance value, and the RGB information acquisition Conversion means for converting the RGB luminance data of each inspection part acquired from the means into luminance data of the polar coordinate system, and the RGB reference in the RGB polar coordinate system stored in the reference data storage means and the converted luminance data of the polar coordinate system Compared with the luminance data, depending on whether or not the RGB reference luminance data includes the RGB luminance data of the inspection object. It is obtained so as to provide a judging means for judging quality of 査領 zone.

In this way, even if the RGB luminance values in the acquired inspection region vary widely from product to product, as shown in FIG. 3, the overall luminance is maintained while maintaining the RGB luminance ratio (color shade). Only a large value can be set, so that variations in luminance can be absorbed and inspection can be performed with high accuracy.

Further, in such an invention, a conversion table for converting the luminance data of the RGB orthogonal coordinate system acquired from the inspection object into the RGB polar coordinate system is stored, and each of the acquired inspection regions is obtained with reference to the conversion table. Convert RGB coordinates.

In this way, since coordinate conversion is performed with reference to a conversion table prepared in advance, there is no need to perform coordinate conversion every time inspection is performed, and the processing speed during inspection can be increased.

Further, when setting the RGB reference luminance data, RGB luminance data of a predetermined number of pixels adjacent to the reference inspection object is included.

In this way, it is possible to inspect without aligning the reference data and the inspection object in pixel units, and it is possible to inspect with reduced false alarms.

In addition, when RGB luminance is converted into polar coordinate luminance, the luminance value is compressed or expanded so as to fit in one byte.

Usually, in the rectangular coordinate system, the luminance value of RGB falls within the range of 1 byte of “0 to 255”, but when this is converted into the polar coordinate system, the angle (θ, ρ) from the axial direction is from 0. The distance (L) from the origin falls within the range of 0 to 255 × 3 ^(1/2) and exceeds 1 byte. Therefore, the angle from the axial direction is (255 × 4 / π) times, and the distance from the origin is reduced to 1/3 ^(1/2) times. In this way, data can be compressed within a range of 1 byte.

In the present invention, in the appearance inspection apparatus that inspects the formation state of the inspection object based on the image acquired from the inspection object, a color image is acquired from the inspection object, and the RGB luminance in the inspection portion of the inspection object RGB information acquisition means for acquiring data, reference data storage means for storing RGB reference luminance data of each inspection part in an RGB polar coordinate system with the axial direction as a luminance value, and each inspection part acquired from the RGB information acquisition means Converting the RGB luminance data of the RGB coordinate data into luminance data of the polar coordinate system, comparing the converted luminance data of the polar coordinate system with the RGB reference luminance data in the RGB polar coordinate system stored in the reference data storage unit, The quality of the inspection area is determined based on whether or not the RGB reference luminance data includes the RGB luminance data of the inspection object. Since the RGB luminance values in the acquired inspection area vary greatly from product to product, only the overall luminance is increased while maintaining the RGB luminance ratio (color shade). Therefore, it is possible to absorb the variation in luminance and to inspect with high accuracy.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows an outline of the appearance inspection process in the present embodiment, and FIG. 2 shows a functional block diagram of the appearance inspection apparatus 1. FIG. 3 shows RGB luminance reference data used in the appearance inspection apparatus 1.

The appearance inspection apparatus 1 in this embodiment can inspect a printed circuit board, a semiconductor wafer, other cracks generated in an article, a state printed on the surface of the article, and the like. In this embodiment, an example is given. The case where the formation state of the printed circuit board is inspected will be described.

As shown in the functional block diagram of FIG. 2, the appearance inspection apparatus 1 is acquired by the imaging means 2 that acquires a surface image from the inspection object 11 and the imaging means 2, as in a general appearance inspection apparatus. RGB information processing means for obtaining RGB information of each pixel from the obtained image, correction processing means 6 for aligning the image of the inspection object 11 and the image that is the reference data, and the inspection object subjected to the alignment correction in this way And determining means 8 for determining the quality of the pixel using the RGB luminance data of each pixel of the object 11. Characteristically, the conversion means 7 for converting the RGB luminance data of each image acquired from the inspection object 11 into the RGB luminance data of the polar coordinate system, and the converted RGB luminance data and stored in the storage means 5 in advance. This is compared with the RGB luminance reference data of the polar coordinate system, and the quality of each pixel is judged, and the result can be outputted via the output means 9. Hereinafter, a specific configuration of the appearance inspection apparatus 1 will be described in detail.

First, the imaging means 2 acquires the surface image from the reference object 10 and the inspection object 11 necessary for the inspection, and acquires the surface image by color. This imaging means 2 irradiates light from an oblique direction, and acquires the reflected light by the CCD camera or the like above it. At this time, images are acquired using different angles, different colors, and brightness with respect to the reference object 10 and the inspection object 11, and the acquired images are selected and used. The reference object 10 generates reference data as a reference when the inspection object 11 is inspected. Generally, the reference object 10 that has been determined to be non-defective by visual inspection or other inspection devices is generally used. To do.

The reference data generation means 4 acquires a surface image from a reference object 10 prepared in advance, and generates reference data from the image of the reference object 10. The generated reference data includes data relating to the entire shape of the reference object 10, data relating to a plurality of rectangular areas inside the reference object 10, data relating to each pixel, and the like. Of these, data relating to the overall shape includes data relating to the length and width of the printed circuit board, data relating to the rectangular area includes data such as pattern images in the rectangular area, and data relating to each pixel includes data relating to each pixel. Data such as RGB luminance, allowable luminance width, and search distance are used. Here, the “allowable luminance width” indicates the RGB luminance width in a pixel that is determined to be good or bad. For example, a portion having a large luminance change such as a silk edge, a pad edge, or a wiring pattern edge. Is set to be large. The “search distance” indicates a distance for searching whether or not there is a pixel corresponding to the reference object 10 with a predetermined pixel position as the center. For example, a silk edge, a pad The search distance is set to a large value such as 3 to 5 pixels for a portion where the luminance change is large such as the edge of the wiring pattern or the edge of the wiring pattern. These allowable luminance width and search distance are automatically set, and the upper limit value of the allowable luminance width and search distance is set manually in advance. For example, in FIG. 1, the allowable luminance width expressed in polar coordinates is ± Δθ, ± Δρ, ± ΔL are set, and the search distance is set to 3 pixels or the like. The allowable luminance width and the search distance are not limited to these values, and may be set manually.

Figure 3 shows an overview of this reference data. FIG. 3 shows reference data used when determining the quality of each pixel. The luminance values of RGB are set on the respective axes, and the luminance values are set to increase in the direction of the arrows. In general, when the reference data is set in such an RGB orthogonal coordinate system, the RGB hue slightly changes depending on the resist unevenness of the inspection object 11, the change of the lot, and the color composition. In particular, when the luminance change is large depending on the product, the luminance width from dark green to bright green changes within the range of (R, G, B) = (20, 120, 60) to (40, 200, 150). It may end up. For this reason, if the reference luminance data is set in the RGB orthogonal coordinate system, the luminance width must be set as large as 20 <R <40, 120 <G <200, 60 <B <150 (FIG. 9). State). However, if the luminance width of the RGB luminance reference data is set to be large as described above, most pixels are included in the range of the reference data. Anything that can be seen is included in the range of the reference data and judged as “good”. More specifically, when there is a portion where the resist is peeled off and the copper of the pattern is exposed, the portion must be determined as “bad” originally, but this portion is partially Since it is covered with a thin resist, it often has a slightly reddish resist color (luminance values (R, G, B) = (35, 150, 80)). For this reason, if such a portion is determined based on the reference data of the RGB orthogonal coordinate system, it cannot be determined as defective.

Therefore, in this embodiment, as shown in FIG. 3, RGB luminance reference data expressed in the RGB polar coordinate system space is used. If such a coordinate system is used, even if the luminance value is reduced as a whole while maintaining the balance of the RGB luminance values, the pixel can be determined as a good pixel. That is, as shown in FIG. 3, when determining pixels within the range of luminance values (R, G, B) = (20, 120, 60) to (40, 200, 150), the entire RGB luminance value is determined. Thus, it is possible to determine a pixel that is different from the entire luminance value while maintaining the above balance as a good pixel. On the other hand, pixels with different brightness value balances such as (R, G, B) = (35, 150, 80), that is, pixels with completely different hues can be determined as defective pixels. In addition, when generating the RGB luminance reference data of these pixels, the RGB luminance values of the pixels adjacent to the pixel are set to be included. Then, the RGB luminance reference data set in this way is stored in the storage means 5.

The correction processing unit 6 performs correction processing for making the image of the inspection object 11 imaged by the imaging unit 2 substantially coincide with the image of the reference object 10. An example of the correction of the entire image in this correction processing is shown in FIG. In FIG. 4, a solid line portion shaded with diagonal lines indicates the reference object 10, and a broken line indicates the inspection object 11. As shown in FIG. 4, when the inspection object 11 is smaller than the reference object 10 (FIG. 4A), correction processing is performed to enlarge the entire shape by δ _x and δ _y . Also, if the inspection object 11 is rotated by also [delta] _theta than reference object 10 performs correction processing as rotated by that angle. Further, when the inspection object 11 is displaced in parallel with the reference object 10, a correction process is performed in which the inspection object 11 is translated by the amount of the displacement. These correction processes are effective when, for example, the inspection object 11 is not fixed at a regular position on the stage, or when there is an error in the dimensions of the inspection object 11.

Next, another aspect of this correction processing is shown in FIG. FIG. 5A shows an example of a rectangular area where the reference object 10 is located, and FIG. 5B shows a rectangular area at the same position of the inspection object 11. In an actual product, as shown in FIG. 5B, the pad or wiring pattern of the inspection object 11 may be displaced in a predetermined direction from the pad or wiring pattern of the reference object 10. In such a case, a parallel movement correction process is performed so that the image in the rectangular area of the inspection object 11 substantially matches the reference object 10. By performing these correction processes, the pad and the wiring pattern of the inspection object 11 substantially coincide with the pad and the wiring pattern of the reference object 10, and the allowable luminance can be obtained without increasing the search distance too much. Pixels within the width can be found. That is, originally, it is possible to find a corresponding pixel with a narrow search distance, where the search distance must be extended to a corresponding position in the image that has been corrected before correction. By narrowing the search distance in this way, it is possible to prevent an irrelevant pixel that happens to have the same luminance to be determined as a “corresponding pixel”.

Then, the RGB information acquisition means 3 acquires RGB luminance data of each pixel from the image of the inspection object 11 that has been corrected in this way. At this time, the acquired RGB luminance data is information included in the range of (R, G, B) = (0, 0, 0) to (255, 255, 255), but is expressed in this orthogonal coordinate system. The converted coordinates are converted into coordinates in the RGB polar coordinate system using the conversion means 7.

In the case where the orthogonal coordinate system is converted into the polar coordinate system using the conversion means 7, in this embodiment, as shown in FIG. 6, the values (x, y, z) of the orthogonal coordinate system are previously set to the values of the polar coordinate system (θ , Ρ, L), a conversion table to be converted is prepared, and converted to RGB luminance data of the polar coordinate system with reference to this conversion table. In general, the relationship between orthogonal coordinates and polar coordinates is as follows.
x = Lsinρcosθ
y = Lsinρsinθ
z = Lcosθ
Where x is the luminance value of R in the Cartesian coordinate system, y is the luminance value of G in the Cartesian coordinate system, z is the luminance value of B in the Cartesian coordinate system, θ is the angle made with the x axis in the polar coordinate system, and ρ is the polar coordinate The angle L formed with the z-axis in the system indicates the overall luminance value in the polar coordinate system.

Therefore, these relationships are inversely converted to convert the value of (x, y, z) into the following polar coordinate system values, and these converted values are stored in the storage means 5 as a conversion table.
ρ = tan ⁻¹ {(x ² + y ² ) ^1/2 / z}
θ = tan ⁻¹ (y / x)
L = (x ² + y ² + z ² ) ^1/2

By the way, when the luminance values (0 to 255) in the orthogonal coordinate system are converted into the luminance values in the polar coordinate system, θ and ρ are 0 ≦ θ ≦ π / 2 and 0 ≦ ρ ≦ π / 2. , L, 0 ≦ L ≦ 255 × 3 ^(1/2) . Therefore, θ and ρ are smaller than 1 byte, and L exceeds 1 byte. Therefore, the converted luminance value is multiplied by 255 × 2 / π so that θ and ρ are exactly 1 byte, and for L, the converted luminance value is 1/3 so that it is exactly 1 byte. Compress to ^(1/2) times.

The determination means 8 determines whether or not the inspection object 11 has a pixel corresponding to each pixel of the reference object 10, and the first pixel determination means 81 and the second pixel as described below. A pixel determination unit 80 including a determination unit 82 and a cluster determination unit 83 are provided.

First, the first pixel determination unit 81 specifies the position of the inspection object 11 corresponding to each pixel of the reference object 10 with the reference object 10 as a reference, and within a search distance centered on this position. Then, it is determined whether or not there is a pixel within an allowable luminance width with respect to the luminance of the pixel. In this determination, if even one pixel within the allowable luminance width of the RGB luminance reference data exists within the search distance, it is determined as “good pixel”, and conversely, a pixel within the allowable luminance width within the search distance. Is not determined at all, it is determined as “defective pixel”. Normally, if the correction processing means 6 can completely match the reference object 10 and the inspection object 11, the pixel at the position of the inspection object 11 corresponding to the pixel position of the reference object 10 may be inspected. However, in reality, there are optical shifts and mechanical shifts, so that it is difficult to make them completely coincide with each other. Further, when the resolution is increased, there is a possibility that the shift is about several pixels. For this reason, if there is an almost identical luminance within the search distance, it is determined as “good pixel” as a primary determination. The determination result by the first pixel determination unit 81 is visually displayed on a display device or the like. For example, in the portion determined as “defective pixel”, an “x” mark or the like is displayed on the image of the reference object 10. Indicates.

On the contrary, the second pixel determination means 82, on the contrary, uses the inspection object 11 as a reference, and within the search distance centered on the position of the reference object 10, the allowable luminance width for the luminance of the pixel of the inspection object 11 It is determined whether or not there is a pixel inside. Also in this determination, if there is even one pixel within the allowable luminance width within the search distance, it is determined as “good pixel”, and conversely, there is no pixel within the allowable luminance width within the search distance. Is determined as “defective pixel”. When the comparison process is performed using the inspection object 11 as a reference, an image of the inspection object 11 after the above correction process is used. Then, the position of the reference object 10 corresponding to the position (the center position of the first search distance) of the inspection object 11 after the correction processing is specified, and the allowable luminance width / search distance at the position is stored. It is read out from the means 5 and it is determined whether or not a pixel within the allowable luminance width with respect to the luminance at the position of the inspection object 11 exists on the reference object 10 within the search distance. In this case, in the relationship with the present invention, the first search distance and the second search distance match, and the first allowable luminance width and the second allowable luminance width match. Normally, if the correction processing means 6 can completely match the reference object 10 and the inspection object 11, the pixel at the position of the inspection object 11 corresponding to the pixel position of the reference object 10 may be inspected. However, in reality, there are optical shifts and mechanical shifts, so that it is difficult to make them completely coincide with each other. Further, when the resolution is increased, there is a possibility that the shift is about several pixels. For this reason, if there is a pixel having substantially the same brightness within the search distance, it is determined as “good pixel” as a secondary determination. Similar to the first determination result, the second determination result is visually displayed on the display device, overwritten on the determination image by the first pixel determination unit 81, and determined as a “defective pixel”. An “x” mark or the like is shown in the part.

Finally, the pixel determination unit 80 determines that there is at least one pixel within the allowable luminance width within the search distance of the inspection object 11, and within the allowable luminance width within the search distance of the reference object 10. On the condition that at least one pixel exists, the pixel existing at the position of the reference object 10 is determined as a good pixel. Conversely, when there is no pixel within the allowable luminance width within the search distance of the inspection object 11, or when there are no pixels within the allowable luminance width within the search distance of the reference object 10, A pixel present at the position of the reference object 10 is determined as a defective pixel.

Based on the size of the pixel group of the reference object 10 determined as “defective pixel” by the pixel determining means 80, the cluster determining means 83 determines whether or not the inspection object 11 is a defective product as a whole. judge. This pass / fail determination is determined to be a defective product when there are a predetermined number or more of adjacent pixels determined to be “defective pixels”.

The output unit 9 outputs the determination result by the cluster determination unit 83 in a reportable manner. At this time, since it is necessary to inform the user which part is a defective cluster, the position of the cluster determined to be a defective cluster by the cluster determination means 83 is visually output to the display device.

Next, a method for inspecting the inspection object 11 using the appearance inspection apparatus 1 configured as described above will be described.

<Standard data generation flow>
First, FIG. 7 shows a flowchart for generating reference data when inspecting the inspection object 11. When generating reference data, first, each image is acquired from a plurality of reference objects 10 prepared in advance (step S1). When a predetermined number or more of images of the reference object 10 are captured, the data relating to the entire area, the data relating to the rectangular area, and the data relating to the pixels are generated for each reference object 10 (step S2). For the object 10, an average value of data relating to the entire area, an average value of data relating to the rectangular area, an average value of RGB data relating to pixels, and a standard deviation value are calculated (step S3). Next, the upper limit value of the allowable luminance width and the upper limit value of the search distance are manually input (step S4). This input is not performed at this stage, but may be performed in advance before step S1.

Then, after the calculation of the average value and the standard deviation value in step S3, for the pixels having a large standard deviation value, the upper limit value of the allowable luminance width and the upper limit value of the search distance that are input previously are set, For pixels having a small standard deviation value, the allowable luminance width and the search distance are set to be small (step S5). Then, RGB luminance reference data is generated with RGB polar coordinates for each pixel and stored in the storage means 5 (step S6).

<Inspection flow of inspection object 11>
Next, a flowchart for inspecting the inspection object 11 is shown in FIG. First, when inspecting the inspection object 11, the surface image is acquired from the inspection object 11 (step T1). This captured image may be misaligned depending on the image acquisition method, and may be different from the state of the image of the reference object 10 stored in the storage unit 5. Therefore, correction processing is performed to make the image states substantially coincide (step T2). In this correction process, first, the entire shape is corrected. Specifically, three corner points on the inspection object 11 are extracted, and the vertical and horizontal lengths, rotation angles, parallel movement distances, and the like of the inspection object 11 are calculated from the three points. Then, based on these vertical and horizontal lengths, rotation angles, parallel movement distances, etc., correction processing is performed so that the entire image of the inspection object 11 substantially matches the entire image of the reference data.

Next, the rectangular area is corrected. When this rectangular area correction processing is performed, the image of the inspection object 11 is translated so that the image of the predetermined rectangular area of the reference object 10 and the image of the corresponding rectangular area of the inspection object 11 substantially coincide. .

Then, after these correction processes are completed, it is determined whether or not the pixels present at each position of the reference object 10 are good pixels. In this determination, first, the position, RGB luminance, allowable luminance width, and search distance for each pixel of the reference object 10 are read from the storage means 5 (step T3). Then, the position of the inspection object 11 corresponding to the read pixel is specified, and within the RGB luminance reference data (allowable luminance width) set in the RGB polar coordinate system within the search distance with the position as the center. It is determined whether or not there is a pixel (step T4). When using the RGB luminance reference data set in the RGB polar coordinate system, the RGB luminance data of the pixel read from the inspection object 11 is converted into the polar coordinate system with reference to the conversion table, and the RGB luminance data in the polar coordinate system is converted. And compare. When the first pixel determining unit 81 determines that no pixel within the allowable luminance width exists within the search distance, the pixel at the position of the reference object 10 is determined as a “defective pixel”. (Step T8).

Next, after the first pixel determination is completed, the position of the reference object 10 corresponding to the position of the inspection object 11 is determined based on the image of the inspection object 11 after the correction process. It is determined whether or not there is a pixel within the RGB luminance reference data (allowable luminance width) set by the RGB polar coordinates within the search distance as the center. However, since each pixel of the inspection object 11 does not store data such as an allowable luminance width and a search distance, the allowable luminance width and the search of the position of the reference object 10 corresponding to the position of the inspection object 11 are stored. Use distance.

Specifically, first, an allowable luminance width and a search distance of the position of the reference object 10 corresponding to the position of the inspection object 11 are read (step T5). Then, based on the read allowable luminance width and the search distance, the tolerance for the RGB luminance data of the inspection object 11 expressed in the polar coordinate system within the search distance with the position of the corresponding reference object 10 as the center. It is determined whether or not there is a pixel within the luminance width (step T6). At this time, the determination is made by converting the RGB luminance data acquired from the inspection object 11 into the RGB luminance data of the polar coordinate system using the conversion table. If it is determined by the second pixel determining means 82 that there is no luminance pixel within the allowable luminance width within the search distance, the pixel at the position of the reference object 10 is determined as a “defective pixel”. (Step T8).

On the other hand, if it is determined as “good pixel” in step T4 and “good pixel” is determined in step T6, the pixel corresponding to the position of the reference object 10 is determined as “good pixel” (step T7). .

If all the pixels have been inspected (step T9; Yes), the number of adjacent defective pixels among the pixels of the reference object 10 determined as “defective pixels” by the pixel determination unit 80 is then determined. When there are more than a predetermined number of defective pixels (step T10), the inspection object 11 outputs that it is a defective product (step T11), while the number of all adjacent defective pixels Is less than the predetermined number, an output indicating that the product is non-defective is output (step T12).

As described above, according to the embodiment, in the appearance inspection apparatus 1 that inspects the formation state of the inspection object 11 based on the image acquired from the inspection object 11, a color image is acquired from the inspection object 11. RGB information acquisition means 3 for acquiring RGB luminance data in the inspection portion of the inspection object 11, storage means 5 for storing RGB reference luminance data of each inspection portion in the RGB polar coordinate system with the axial direction as the luminance value, The RGB luminance data of each inspection part acquired from the RGB information acquisition unit 3 is compared with the RGB reference luminance data in the RGB polar coordinate system stored in the storage unit 5, and the inspection object 11 is included in the RGB reference luminance data. Since the determination means 8 for determining whether the inspection area is good or not is provided depending on whether or not the RGB luminance data is included. Even if the RGB luminance values in the inspection area vary greatly from product to product, only the luminance can be set large while maintaining the RGB balance, and the luminance variation can be absorbed and high quality inspection can be performed. become able to.

Further, a conversion table for converting the luminance data of the RGB orthogonal coordinate system acquired from the inspection object 11 into the RGB polar coordinate system is stored, and the RGB coordinates of each acquired inspection area are converted using this conversion table. Since each pixel is inspected, it is not necessary to perform an operation for coordinate conversion, and the processing speed at the time of inspection can be increased.

Further, when setting the RGB reference luminance data, the RGB luminance data of a predetermined number of pixels adjacent to the reference inspection object 11 is included, so that the reference data and the inspection object 11 are aligned in pixel units. It can be inspected without it, and it can be inspected with fewer false alarms.

In addition, when the RGB luminance is converted to the polar coordinate luminance, the luminance value is compressed or expanded to fit within 1 byte, so the converted luminance value is stored within the 1-byte range. Can be compressed.

Note that the present invention is not limited to the above-described embodiment, and can be implemented in various modes.

For example, in the above-described embodiment, the correction processing unit 6 is used to correct the image of the inspection object 11, but the image of the reference object 10 may be corrected. Alternatively, such a correction process may not be performed when complete alignment is possible.

In the above embodiment, a conversion table for converting to polar coordinates is prepared in advance, and the conversion table is referred to convert from the orthogonal coordinate system to the polar coordinate system. However, coordinate conversion is performed for each pixel. You may make it calculate.

Furthermore, although the case where the printed circuit board is inspected has been described in the above embodiment, the present invention can also be applied to an inspection object 11 that requires an appearance inspection other than this.

In addition, in the above embodiment, each pixel is inspected, but RGB luminance data may be collected and inspected for each predetermined area. In this case, one RGB luminance data is calculated for each region, and the calculation result is compared with the RGB reference luminance data expressed in the polar coordinate system.

In the above embodiment, θ and ρ are expanded to 255 × 2 / π times. However, since they are within the range of 1 byte, the luminance value may be used as it is.

In the above embodiment, since the RGB element of one pixel is 8 bits, the maximum value is set to 255. However, this is a value of 2 gradation bits raised to −1, and this value can be used depending on the gradation bits. it can.

The figure which shows the outline | summary of the inspection processing method in one embodiment of this invention Functional block diagram of an appearance inspection apparatus in the same form The figure which shows the RGB brightness | luminance reference data of the polar coordinate system in the same form The figure which shows the outline | summary of the process of the whole correction | amendment in the same form The figure which shows the outline | summary of the process of correction | amendment of the rectangular area in the same form The figure which shows the polar coordinate conversion table in the same form The figure which shows the production | generation flow of the reference data in the form The figure which shows the flow of the inspection process in the same form The figure which shows the RGB luminance reference data of the orthogonal system in a prior art example

DESCRIPTION OF SYMBOLS 1 ... Appearance inspection apparatus 2 ... Imaging means 3 ... RGB information acquisition means 4 ... Reference data generation means 5 ... Storage means 6 ... Correction processing means 7 ... Conversion means 8- ..Determination means 80: Pixel determination means 81 ... First pixel determination means 82 ... Second pixel determination means 83 ... Cluster determination means 9 ... Output means 10 ... Reference object Object 11 ... Inspection object

Claims

In the appearance inspection apparatus that inspects the formation state of the inspection object based on the image acquired from the inspection object,
RGB information acquisition means for acquiring a color image from an inspection object and acquiring RGB luminance data in an inspection part of the inspection object;
Reference data storage means for storing RGB reference luminance data of each inspection portion in an RGB polar coordinate system with the axial direction as a luminance value;
Conversion means for converting RGB luminance data of each inspection portion acquired from the RGB information acquisition means into luminance data of a polar coordinate system;
The converted luminance data of the polar coordinate system is compared with the RGB reference luminance data in the RGB polar coordinate system stored in the reference data storage means, and the RGB luminance data of the inspection object is included in the RGB reference luminance data. An appearance inspection apparatus comprising: a determination unit that determines whether the inspection area is good or not based on whether or not the inspection area is present.
The conversion means stores a conversion table for converting luminance data in an RGB orthogonal coordinate system acquired from an inspection object into an RGB polar coordinate system, and refers to the conversion table to obtain the RGB coordinates of each acquired inspection portion. The appearance inspection apparatus according to claim 1, which is to be converted.
The visual inspection apparatus according to claim 1, wherein the RGB reference luminance data is set to include RGB luminance data of a predetermined number of pixels adjacent to a reference inspection object.
2. The appearance inspection apparatus according to claim 1, wherein the conversion unit compresses or expands the luminance value so as to be within one byte when the luminance of RGB is converted into luminance of a polar coordinate system.