JP2012008619A - Edge detection method, edge detection apparatus. and inspection apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an edge detection method for stably detecting an edge.SOLUTION: A processing unit 12 calculates deviations of pixels having the same luminance value on the basis of the luminance value included in interpolation image data, and calculates the luminance value of the pixel having the smallest deviation. The processing unit 12 defines the calculated luminance value as an edge determination value, and detects, as an edge, a pixel included in the interpolation image data and having a value same as the edge determination value.

Description

  The present invention relates to an edge detection method, an edge detection device, and an inspection device.

  An inspection apparatus for inspecting a printed circuit board inspects the printed circuit board using image data obtained by imaging the printed circuit board with an imaging device such as a CCD (Charge Coupled Device). One of the inspections is an inspection related to the wiring formed on the printed circuit board. The inspection apparatus performs image processing on the image data to detect edges (see, for example, Patent Document 1), and recognizes between the edges as wiring. Then, the wiring width of the wiring is measured, and the quality of the printed circuit board is determined based on whether or not the wiring width is within a specified range.

  In recent years, for example, a printed circuit board on which a flip chip is mounted has been improved in the definition of wiring. When such wiring is measured, the resolution of an image sensor such as a CCD may be insufficient. For this reason, image data having a finer pitch than the pitch of an image sensor such as a CCD is created by interpolating pixel values (for example, luminance) of image data obtained by the imaging device. In this method, the apparent resolution can be increased, so that high-definition wiring can be verified.

  However, in the image data obtained by interpolation, the edge position to be determined differs depending on the setting of the threshold level, and the measurement result, that is, the line width changes greatly. In order to suppress such a change, there are a method for setting the level, a method using an average value, a method using a luminance gradient, and the like.

JP 2005-242596 A

  However, in the method using the average value, a region for obtaining the average value must be set so that the wiring is included. In addition, the size of the area for which the average value is obtained must be set so wide that there is no influence of wiring. However, if the region is too wide, the average value is likely to be affected by uneven illumination.

  In addition, in the method using the luminance gradient, the edge is determined using the gradient (difference in luminance value) in a minute section (between pixels created by interpolation). The slope is calculated. Further, when the amount of light changes depending on the lighting condition, the inclination changes, so that it is difficult to detect an edge.

  The present invention has been made to solve the above problems, and an object of the present invention is to provide an edge detection method, an edge detection device, and an inspection device capable of stably detecting an edge. .

  According to one aspect of the present invention, the brightness of a pixel that has the same brightness value is calculated based on the values of the pixels that form the image, and the brightness value of the pixel that has the smallest shake width is calculated. And a pattern edge detection process in which a calculated luminance value is used as an edge determination value, and a pixel having a value equal to the edge determination value among pixels included in the image is detected as an edge.

  According to the angle between the pixel arrangement direction and the edge included in the interpolated image data, the pixel is swung in a direction orthogonal to the direction in which the pixels having the same luminance value are arranged. This runout width increases as the distance from the edge increases. Therefore, by calculating the luminance value of the pixel having the smallest fluctuation width, the luminance value indicating the edge is obtained and the edge is detected.

  According to one aspect of the present invention, it is possible to provide an edge detection method, an edge detection apparatus, and an inspection apparatus that can detect edges stably.

The schematic block diagram of the test | inspection apparatus of one Embodiment. The flowchart of an edge detection process. Explanatory drawing of image division. (A) is a partially enlarged view of an inspection object, and (b) is an explanatory diagram of image data. Explanatory drawing of an interpolation process. Explanatory drawing which shows the image data after an interpolation process. Explanatory drawing of an edge detection process. Explanatory drawing of an edge detection process. Explanatory drawing of another edge detection process. Explanatory drawing of another edge detection process.

Hereinafter, an embodiment will be described with reference to the accompanying drawings.
As illustrated in FIG. 1, the inspection apparatus includes an imaging unit 11, a processing unit 12, and an output unit 13. The imaging unit 11 is a one-dimensional imaging device (so-called line CCD) including, for example, a CCD (Charge Coupled Device). The imaging unit 11 may be a CMOS imaging device. Further, a two-dimensional image sensor may be used.

  The imaging unit 11 is arranged so as to photograph the inspection object 21. The inspection object 21 is a printed wiring board, for example, and is transported in the transport direction (arrow direction shown in the figure) by the transport device 14 (for example, a belt conveyor).

  The imaging unit 11 is arranged so as to image the entire inspection object 21. For example, the one-dimensional imaging unit 11 is arranged such that elements are arranged along a direction orthogonal to the transport direction on a horizontal plane. Then, the imaging unit 11 outputs image data for one line. The inspection object 21 is conveyed. Therefore, the image data output from the imaging unit 11 is data for each line obtained by dividing the inspection object 21 at a constant interval in the conveyance direction.

  The processing unit 12 has an image memory, and sequentially stores the image data output from the imaging unit 11 in the image memory. Thereby, the image data of the entire inspection object 21 is stored in the image memory.

  The processing unit 12 interpolates the image data stored in the image memory to generate interpolated image data. The interpolation process is a process of increasing the resolution by apparently reducing the pixel interval, that is, the pixel pitch, by inserting the interpolation pixel generated based on the pixel value between the pixels of the image data. The processing unit 12 generates the pixel value of the interpolation pixel by, for example, bicubic interpolation (Bicubic interpolation).

  Next, the processing unit 12 detects the edge of the measurement object (for example, a wiring formed on the substrate) included in the inspection object 21 based on the interpolation image data. Next, the processing unit 12 calculates the distance between the detected edges. The distance between the edges corresponds to the width of the wiring (pattern width). That is, the processing unit 12 extracts the wiring edge from the interpolation image data, and calculates the wiring width.

  Next, the processing unit 12 compares the calculated wiring width with a set value. The set value is a determination value for determining pass / fail of the inspection object 21 and includes a value indicating an allowable range of the wiring width, that is, a minimum wiring width and a maximum wiring width. Accordingly, the processing unit 12 compares the wiring width with the set values (minimum wiring width, maximum wiring width), and determines that the wiring with the wiring width is acceptable (OK) when the wiring width is within the allowable range. When the wiring width exceeds the allowable range, the wiring is determined to be rejected (NG). The processing unit 12 determines that the inspection object 21 is acceptable when all the wirings pass in the entire inspection object 21, and rejects the inspection object 21 including at least one unacceptable wiring. judge. Accordingly, the processing unit 12 includes an interpolation processing unit, an edge detection unit, and a determination unit.

  The processing unit 12 displays various data in the processing on the output unit 13. The output unit 13 is a display device such as an LCD. For example, the processing unit 12 displays an image based on the image data, an image based on the interpolated image data, an image of the block being processed, a determination result for each block, a determination result for the inspection target 21, and the like on the output unit 13. The output unit 13 may be a printing apparatus (printer or the like). In addition, a display device and a printing device may be provided as the output unit 13. In addition, it is good also as a structure provided with a keyboard and a mouse | mouth as an apparatus which inputs the setting value and the instruction | indication with respect to the process part 12. FIG.

Next, processing in the processing unit 12 will be described.
First, the interpolation process will be described.
As shown in FIG. 4A, a wiring 23 having a predetermined width is formed on an insulator 22 in the inspection object 21. The insulator 22 is, for example, an epoxy resin type, a phenol resin type, a polyimide resin type, an unsaturated polyester resin type, or a polyphenylene ether resin type alone, a modified product, or a mixture. The insulator 22 and the wiring 23 have different reflectances. For example, the reflectance of the wiring 23 is higher than that of the insulator 22. Therefore, the image data (luminance value) output from the pixel that images the wiring 23 has a value larger than the luminance value for the insulator 22.

  Furthermore, each pixel of the imaging unit 11 has a predetermined size. Therefore, the image data (luminance value) output from the pixel obtained by imaging the edge 23a of the wiring 23 is based on the reflected light from the portion of the wiring 23 and the portion of the insulator 22 included in the region imaged by the pixel. Therefore, as shown in FIG. 4B, the image data is composed of one pixel, a region 31 corresponding to a pixel obtained by photographing only the insulator 22, a region 32 corresponding to a pixel obtained by photographing only the wiring 23, and one pixel. A region 33 in which the insulator 22 and the wiring 23 are photographed is included. The dotted line shown in FIG. 4B indicates the edge 23a of the wiring 23 shown in FIG. The boundary line between the regions 31 and 32, that is, the boundary line of the luminance value with respect to the insulator 22 and the boundary line of the luminance value with respect to the wiring 23 each have a stepped shape on both sides of the edge 23a. That is, the boundary between the luminance values changes in a staircase pattern.

  A region 33 between the two regions 31 and 32 is data of a luminance value corresponding to the amount of reflected light from the wiring 23 and the insulator 22 in the pixel region. In addition, the area | region 34 imaged by one pixel is shown on the upper left of FIG.4 (b). Note that although the luminance values in the above-described areas 31 and 32 are constant values in principle, the pixel values (luminance values) of actual image data are the states of the surfaces of the wirings 23 and the insulators 22 (for example, unevenness). It changes due to the change of the reflectance based on.

  The processing unit 12 performs an interpolation process on the image data obtained as described above. As described above, the interpolation processing calculates the pixel value of the interpolated pixel by cubic interpolation (Bicubic interpolation). The processing unit 12 divides the adjacent two pixels into 10 in the image data. That is, the processing unit 12 divides an area having four pixels as rectangular vertices into 100 auxiliary areas, and uses the vertices of each auxiliary area as coordinate values of the auxiliary pixels. Then, the processing unit 12 calculates the pixel values of the pixels and the interpolation pixels included in the original image data based on a plurality of surrounding pixel values. FIG. 5 shows pixel values (luminance values) of image data and pixel values (luminance values) obtained by the interpolation process. In FIG. 5, black squares indicate pixel values (measurement points) of image data output from the imaging unit 11, and alternate long and short dash lines indicate line segments connecting pixel values included in the interpolated image data.

  The interpolated image data generated by the above interpolation processing is shown in FIG. FIG. 6 shows the distribution of pixel values (luminance values) of pixels included in the interpolated image data as regions 41 to 45. A solid line in the figure indicates a boundary line of a region in which pixel values are divided into ranges of predetermined values (for example, 20). For example, the region 41 is a region including pixels with luminance values [40] to [60], and the region 42 is a region including pixels with luminance values [60] to [80].

  A boundary between regions including pixels having the same pixel value (luminance value) is obtained, for example, as a line segment connecting pixels having different (for example, small) pixel values of adjacent pixels among pixels included in the region. This boundary line undulates in a direction substantially perpendicular to the direction in which the wiring extends in accordance with the luminance value. That is, in a line connecting interpolation pixels having the same value, a line for a predetermined luminance value is undulated, and a line for another luminance value is not undulated. This wavy phenomenon is caused by the change in the amount of light incident on the pixels of the imaging unit 11 according to the inclination of the wiring 23 (edge 23a) with respect to the arrangement direction (X axis and Y axis) of the pixels constituting the image data. (Strength).

  Reflected light from the edge portion is a bright side (a portion with a relatively large amount of reflected light and a portion of the wiring 23) and a dark portion (a portion with a relatively small amount of reflected light and a portion of the insulator 22). Causes blur on both sides. The amount varies with a Gaussian distribution. Therefore, the amplitude of the line segment connecting the interpolation pixels having the same luminance value gradually decreases from the central portion of the wiring 23 toward the edge 23a. Further, it gradually decreases from the insulator 22 toward the wiring 23, that is, toward the edge 23a. Accordingly, among the line segments connecting pixels having the same luminance value, the line segment having the smallest undulation, that is, the smallest amplitude is the position corresponding to the edge of the wiring 23.

Next, an example of edge detection processing will be described.
The processing unit 12 executes each process according to the flowchart shown in FIG. 2 and detects an edge.

  That is, in step 51, the processing unit 12 inputs interpolation image data (step 51: image input processing), and divides the image into a plurality of rectangular areas (blocks) (step 52: block division processing). For example, as illustrated in FIG. 3, the processing unit 12 generates nine blocks 61 by dividing the interpolated image data 60 into three parts each in vertical and horizontal directions. Note that the number of divisions may be appropriately changed to a value of 1 or more. Moreover, you may set according to the magnitude | size of the test target object 21, the state of the illumination with respect to the test target object 21, etc. Further, the number of divisions may be changed according to the state of image data, for example, the state of luminance.

  Next, the processing unit 12 extracts a diagonally extending pattern (wiring) based on the pixel value (step 53: diagonal pattern extraction process). As described with reference to FIGS. 4 to 6, the wiring extending obliquely with respect to the image data arrangement direction (the pixel arrangement direction, that is, the X axis and the Y axis) is a line segment that connects pixels having the same luminance value. Give rise to Accordingly, the processing unit 12 detects a line segment connecting pixels having the same luminance value for a predetermined pixel value (luminance value), and detects a line segment extending obliquely depending on whether the line segment is wavy. Note that the undulation of the line connecting the pixels can be detected by periodically changing the increase amount (or decrease amount) of the coordinate value of the pixel. For a line segment that is not wavy, that is, a straight line, the increase amount (decrease amount) of the coordinate value of the pixel (for example, the coordinate value of the X axis) is constant.

  Then, the processing unit 12 calculates the extending direction of the extracted pattern (wiring). This direction is obtained by an approximate straight line, for example. That is, the processing unit 12 calculates a coefficient (a, b) of a linear equation (y = ax + b) of an approximate line using each coordinate value as a parameter based on the position (coordinate value) of the pixel having the same luminance value. The coefficient (a) indicates the direction (inclination) in which the wiring extends.

Next, the processing unit 12 (step 54: luminance calculation processing). The processing unit 12 obtains luminance as an edge.
As shown in FIG. 7, wavy line segments generated at two parallel edges of the wiring are generated symmetrically with respect to the center of the wiring. In the figure, the arrow indicates the direction in which the wiring extends. Two wavy line segments 71a. 71b (indicated by a one-dot chain line in the figure) intersects with a line segment extending in a direction orthogonal to this direction. The processing unit 12 forms a line segment with a predetermined number of pixels along the direction of the pattern detected in step 53. At this time, the number of pixels used for determining the line segment may be any number that can determine the undulation of the line connecting the pixels, that is, the amplitude of the line segment. For example, the number corresponding to one cycle of the undulation (For example, 50). The processing unit 12 calculates an interval (inter-intersection distance) between the intersections 72a and 72b.

  As shown in FIG. 8, the line segment 71 indicating the distance between the intersections increases or decreases with respect to the position in the wiring extending direction (for example, the value when the coordinate value of the X-axis is coordinate-converted in the extending direction of the wiring). repeat. The line segment with the smallest increase / decrease is the line segment with the smallest waviness (amplitude). In the line segment shown in FIG. 7, the line segments 73a and 73b (the line segment 73 in FIG. 8) are the line segments with the smallest waviness. Then, the processing unit 12 obtains the luminance of this line segment. Note that the line segments shown in FIGS. 7 and 8 are line segments for each predetermined value (for example, 20). For this reason, there is a case where a line segment obtained by calculation with the least waviness is not shown.

  Next, the processing unit 12 (step 55: pattern edge detection processing). The processing unit uses the obtained luminance value as an edge determination value, and detects an edge of the pattern included in the block 61 being processed based on the edge determination value. The edges detected at this time include the edges of the entire diagonal pattern detected in step 53 and the edges of other patterns (for example, the upper left and lower right edges 23a in FIG. 4A. Further, the detected edges. Includes edges that are parallel to the pixel arrangement direction, and it is difficult to detect edges even at edges parallel to the pixel arrangement direction due to optically generated blur. By detecting the edge based on the calculated luminance value, the edge in the pixel arrangement direction can be detected stably.

  Next, the processing unit 12 determines whether or not an edge has been detected in all the blocks 61 (step 56: end determination processing). When the processing unit 12 detects an edge in all the blocks 61, the processing ends. On the other hand, when there is an undetected block 61, the processing unit 12 proceeds to step 53.

  As described above, the processing unit 12 executes the processing of steps 53 to 55 for each rectangular area (block) 61. Therefore, for each block 61, the processing unit 12 calculates a luminance value representing a line segment with the smallest undulation, and detects an edge based on the luminance value.

  That is, the luminance value for detecting an edge in each block 61 is set according to the pixel value of the image included in the block 61. Therefore, even when luminance unevenness due to illumination or the like occurs on the inspection object 21, the luminance value corresponding to the luminance unevenness is calculated, so that the influence of the luminance unevenness can be reduced.

  The occurrence of luminance unevenness can be determined in advance by a trial inspection (inspection using a test substrate). For this reason, the number of divisions for dividing the interpolated image data 60 into blocks 61 (see FIG. 3) may be set in accordance with the luminance unevenness occurring in the inspection object 21. By setting in this way, it is possible to perform edge detection and the width measurement of the wiring 23 that are not affected by luminance unevenness, and to reduce erroneous determinations for inspection.

  In the above description, the line segment connecting the pixels is described. However, this is for easy understanding of the positional relationship between the pixels having the same luminance value, and the line segment is not necessarily used as an actual process. . For example, the processing unit 12 pays attention in order from the maximum value (or the minimum value) of the luminance values of the pixels included in the block 61, and extracts pixels having the focused luminance value. Then, the wavy amount (amplitude) may be calculated for pixels in a non-linear state that are not linearly arranged based on the coordinate values of the extracted pixels.

As described above, according to the present embodiment, the following effects can be obtained.
(1) Based on the luminance value of the pixel included in the interpolated image data, the processing unit 12 calculates the pixel fluctuation width for a plurality of pixels having the same luminance value, and calculates the luminance value of the pixel having the smallest fluctuation width. . Then, the processing unit 12 uses the calculated luminance value as an edge determination value, and detects a pixel having a value equal to the edge determination value among the pixels included in the interpolated image data as an edge.

  Corrugation occurs in line segments connecting pixels having the same luminance value in accordance with the arrangement direction (X axis and Y axis) of the pixels included in the interpolated image data and the angle formed by the edge 23a of the wiring 23. That is, the position of the pixel having the same luminance value is different from the linear state, that is, the pixel is shaken in a direction orthogonal to the direction in which the pixels having the same luminance value are arranged. This fluctuation width (line segment amplitude) increases as the distance from the edge 23a increases. Therefore, the luminance value indicating the edge can be obtained by calculating the luminance value of the pixel having the smallest fluctuation width. And the edge in the block 61 can be stably detected by this luminance value.

  (2) Waves of the line segment (pixel fluctuation width) occur in the same manner even if the luminance value of the pixel changes. Therefore, it is possible to detect the edge stably without being affected by the change of the luminance value of the pixel, that is, the light hit condition.

  (3) The processing unit 12 divides the interpolated image data 60 into a plurality of blocks 61, and calculates the luminance value of the pixel having the smallest amplitude of the line segment for each block 61. As a result, even if a luminance change for each block 61, for example, luminance unevenness occurs, an edge can be detected stably without being affected by it.

  (4) The processing unit 12 performs an interpolation process on the image data obtained by imaging the inspection object 21 by the imaging unit 11 to generate an interpolation pixel having a pitch smaller than the pitch of the imaging element. Then, the processing unit 12 determines the edge based on the interpolation image data including the interpolation pixel. Thereby, the apparent resolution can be increased and the edge 23a can be detected stably, whereby the width of the high-definition wiring 23 can be measured.

  (5) The processing unit 12 detects an oblique pattern from the interpolated image data, and calculates a luminance value for determining an edge based on a line segment of a wavy phenomenon generated by the oblique pattern. Therefore, the edge can be detected by a smaller number of pixels compared to the entire interpolation image data or the number of pixels necessary for calculating the average value. For this reason, the time required for the edge detection process can be shortened. Further, since the edge can be detected with a small number of pixels, the edge can be detected even with the processing unit 12 having a low processing capability.

In addition, you may implement the said embodiment in the following aspects.
In the above-described embodiment, the luminance value is calculated by detecting a line segment with few undulations based on the distance between the wavy line segments generated by the two parallel edges 23 a in the wiring 23. The present invention is not limited to the above method as long as a luminance value representing a line segment with few undulations can be calculated. For example, as shown in FIG. 9, a line segment with few undulations may be detected for the undulation line segments 81 to 84 generated at one edge.

  For example, each line segment shown in FIG. 9 is rotationally converted as shown in FIG. 10 with respect to one axis (for example, the X axis) of the image data, and the coordinate values of the Y axis for each of the converted line segments 81a to 84a. A line segment with a small change may be detected as a line segment with few wavinesses.

  Further, a straight line 85 is virtually generated in the wiring extending direction (the direction indicated by the arrow in FIG. 9), and the distance between the straight line 85 and each of the line segments 81 to 84 is calculated, and the change in the distance is small. You may make it detect a line segment as a line segment with few undulations.

  In addition, the straight line extending along the direction in which the wiring extends has a uniform coordinate value difference between pixels in the image data coordinate system (X-axis and Y-axis). Therefore, on the basis of the coordinate values of the pixels forming the line segment, a line segment with a small increase / decrease in the coordinate value difference between the pixels is detected as a line segment with less waviness on at least one of the X axis and the Y axis. May be.

  Alternatively, two line segments having the same amplitude may be detected, and a line segment having an intermediate value (average value) of luminance values of these line segments may be detected as a line segment with less waviness. Two line segments having the same amplitude appear symmetrically with respect to the edge.

  The inspection object 21 in the above embodiment may be a high-definition printed circuit board, a lead frame, a semiconductor wafer, a photomask thereof, or the like, or may detect an edge of a pattern formed on them.

In contrast to the above-described embodiment, methods such as linear interpolation, nearest neighbor interpolation (nearest taber), bilinear interpolation (bilinear interpolation), or the like may be used as a method for calculating an interpolated pixel value.
In the above embodiment, the edge 23a is detected by calculating a luminance value with less waviness based on a change in the pixel value (luminance value) generated with respect to the linear edge 23a. On the other hand, you may make it detect the edge formed in arbitrary curves. In an arbitrary curve, for example, in a portion where the tangential direction of the curve and the arrangement direction of the pixels constituting the image data make an angle, a undulation phenomenon occurs similarly to the linear edge 23a. It is also possible to calculate a luminance value as an edge in a portion where the undulation phenomenon occurs and detect an edge of the entire image (or an image in a block) based on the luminance value.

DESCRIPTION OF SYMBOLS 11 Imaging part 12 Processing part 21 Inspection object 23 Wiring 23a Edge 60 Interpolation image data 61 Block 71a, 71b Line segment

Claims (7)

An edge detection method for detecting an edge in an image,
Based on the values of the pixels constituting the image, a luminance calculation process for calculating a pixel shake width for a plurality of pixels having the same brightness value, and calculating a brightness value of a pixel having the smallest shake width;
A pattern edge detection process for detecting, as an edge determination value, the calculated luminance value, and a pixel having a value equal to the edge determination value among the pixels included in the image,
An edge detection method comprising:   The edge detection method according to claim 1, wherein the fluctuation width of the pixel is an amplitude of a line segment connecting a plurality of pixels having the same luminance value. Including block division processing for dividing the image data into a plurality of blocks,
The edge detection method according to claim 1, wherein the edge determination value is calculated for each block in the luminance calculation process.   The pixel constituting the image includes an interpolated pixel between the pixels generated based on the luminance values of the plurality of pixels obtained by the image sensor. The edge detection method described in 1. Including an oblique pattern detection process for detecting an oblique pattern extending along a direction different from the pixel arrangement direction;
In the luminance calculation processing, two line segments connecting a plurality of pixels having the same luminance value are detected, and a distance between two intersections at which the two line segments intersect with a line segment extending in a direction orthogonal to the oblique pattern The edge detection method according to claim 1, further comprising: calculating a luminance value of a line segment with the smallest variation in the distance. An edge detection device including a processing unit for detecting an edge in an image,
The processing unit calculates a pixel shake width for a plurality of pixels having the same luminance value based on a value of a pixel constituting the image, calculates a luminance value of a pixel having the smallest shake width, and calculates the calculated luminance. The value is an edge determination value, and among pixels included in the image, a pixel having a value equal to the edge determination value is detected as an edge.
An edge detection apparatus characterized by the above. An imaging unit for imaging an inspection object;
A processing unit that inspects a pattern width formed on the inspection object based on image data output from the imaging unit;
Including
The processor is
Based on the value of the pixel included in the image data, the pixel shake width is calculated for a plurality of pixels having the same brightness value, the brightness value of the pixel having the smallest shake width is calculated, and the calculated brightness value is subjected to edge determination. As a value, among pixels included in the image, a pixel having a value equal to the edge determination value is detected as an edge, a distance between two edges is calculated as the pattern width, and the pattern width is compared with a set value. To inspect the pattern width,
Inspection apparatus characterized by that.

Images