JP2010118046A - Image processing method, image processor, and surface inspection device using the image processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing method for obtaining an image including an object surface image having an edge area appropriately processed. <P>SOLUTION: The method includes: an image acquisition step (S11) of acquiring a target image including pixel gray-scale values and including an object surface image; an edge image specification step (S13) of specifying, as an edge image, an image of an area which includes an edge line of the object surface image set in the target image and has a predetermined width in a direction orthogonal to the edge line; a conversion step (S14) of converting the edge image into a rectangular image having a width corresponding to the predetermined width; an image processing step (S15) of performing one-dimensional image processing to the rectangular image to successively process the contrast values of a plurality of pixels arranged in the direction orthogonal to the width direction; and a reverse conversion step (S16) of reversely converting the rectangular image subjected to the one-dimensional image processing to the image of the original area to generate a processed edge image. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing method, an image processing apparatus, and a surface inspection apparatus using the image processing apparatus that process an image including an object surface image composed of grayscale values in pixel units.

  There has been proposed a surface inspection apparatus that photographs the surface of an object and inspects the defect on the surface of the object based on the obtained image (see Patent Document 1). In this surface inspection apparatus, the surface of an object to be inspected such as a semiconductor wafer or a liquid crystal glass substrate is scanned by a line sensor, and a two-dimensional surface image of the object to be inspected is generated based on an imaging signal output from the line sensor. Is done. Then, each part of the surface image is processed, and defects (film thickness unevenness, pattern defects, etc.) on the surface of the object to be inspected are detected from the shaded state or the like in the processed surface image of each part.

JP 2007-147441 A

  By the way, as processing for the object surface image performed in the surface inspection apparatus as described above, filter processing can be performed in order to emphasize the density of the defective portion in the object surface image. This filtering process emphasizes the gray value of each pixel of interest in consideration of the gray values of surrounding pixels. That is, the object surface image is subjected to two-dimensional image processing for sequentially processing gray values of a plurality of pixels (a target pixel and surrounding pixels) arranged two-dimensionally.

  However, in the edge portion of the object surface image, the gray value changes abruptly between the inside (edge image) and outside (background image) of the edge line. An appropriate processed image cannot be obtained in the partial area.

  The present invention has been made in view of such circumstances, and provides an image processing method and an image processing apparatus capable of obtaining an image subjected to appropriate image processing in the region of the edge portion of the object surface image. A surface inspection apparatus using the image processing apparatus is also provided.

  An image processing method according to the present invention includes an image acquisition step configured to acquire a processed image including an object surface image, which is configured by grayscale values in pixel units, and an edge line of the object surface image set in the processed image. An edge image specifying step for specifying an image of a region having a predetermined width in a direction orthogonal to the edge line as an edge image, and a conversion for converting the edge image into a rectangular image having a width corresponding to the predetermined width. And an image processing step for performing one-dimensional image processing for sequentially processing the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image.

  With such a configuration, the edge image of the region including the edge line of the object surface image set in the image to be processed and having a predetermined width in the direction orthogonal to the edge line becomes a rectangular image having a width corresponding to the predetermined width. The converted rectangular image is subjected to one-dimensional image processing for sequentially processing the gray values of a plurality of pixels arranged in a direction orthogonal to the width of the rectangular image. In the one-dimensional image processing for the rectangular image, the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image are sequentially processed. Therefore, the rectangular image is a straight line corresponding to the edge line of the object surface image. Are sequentially processed in the direction along the line.

  The image processing method according to the present invention further includes an inverse conversion step of inversely converting the rectangular image subjected to the one-dimensional image processing into an image of an original region to generate a processed edge image. can do.

  With such a configuration, the rectangular image that has been subjected to the one-dimensional image processing is inversely converted into an image of the original region, and a processed edge image is generated. As described above, since the one-dimensional image processing for the rectangular image is performed in the direction along the straight line corresponding to the edge line of the object surface image, the one-dimensional image processed rectangular image is obtained by being inversely transformed. The processed edge image is substantially subjected to image processing along the edge line of the object surface image.

  Further, in the image processing method according to the present invention, the edge image specifying step has a first predetermined width in the inner direction of the object surface image of the edge line, and in the outer direction of the object surface image of the edge line. An image of an area having a second predetermined width can be specified as the edge image.

  With such a configuration, even if an edge line that exactly matches the edge of the object surface image is not set in the image to be processed, an image of a region that always includes the edge of the object surface image is specified as the edge image. Can do.

  Further, in the image processing method according to the present invention, the conversion step includes a position conversion step of acquiring a position in the edge image corresponding to each pixel of the rectangular image, and a gray value of each pixel in the edge image. And determining a gray value of a pixel of the rectangular image corresponding to each position in the acquired edge image.

  With such a configuration, the position in the edge image corresponding to each pixel constituting the rectangular image is acquired, and each position in the acquired edge image is based on the gray value of each pixel in the edge image. The gray value of the pixel of the rectangular image corresponding to is determined.

  When the position in the edge image corresponding to the pixel of the rectangular image matches the pixel position of the edge image, the gray value of the pixel of the rectangular image is set to the gray value of the pixel of the corresponding edge image. I can decide. When the position in the edge image corresponding to the pixel of the rectangular image is shifted from the pixel position of the edge image, the gray value of the pixel of the rectangular image is, for example, in the corresponding edge image. It can be determined by interpolation processing based on the gray value of the surrounding pixels of the position.

  Furthermore, in the image processing method according to the present invention, a conversion table representing a position in the edge image corresponding to each pixel of the rectangular image is created in advance, and the position conversion step uses the conversion table to A position in the edge image corresponding to each pixel of the rectangular image can be obtained.

  With such a configuration, the position in the edge image corresponding to each pixel in the rectangular image is determined using the conversion table, so that the position in the edge image corresponding to each pixel in the rectangular image is faster. Can get to.

  Further, in the image processing method according to the present invention, the inverse conversion step includes a position reverse conversion step of acquiring a position in a rectangular image corresponding to each pixel of the image of the original region, and the one-dimensional image processing. Based on the gray value of each pixel in the obtained rectangular image, the gray value of the pixel in the original area image corresponding to each position in the obtained rectangular image is determined, and the processed edge image is generated. And a step.

  With such a configuration, the position in the rectangular image corresponding to each pixel constituting the image of the original area is acquired, and the acquired rectangle is based on the gray value of each pixel in the one-dimensionally processed rectangular image. The gray value of the pixel of the image of the original area corresponding to each position in the image is determined. A processed edge image constituted by the gray value of each pixel of the image of the original area determined in this way is generated.

  When the position in the rectangular image corresponding to the pixel of the original region image matches the pixel position of the rectangular image, the gray value of the pixel of the original region is the gray value of the pixel of the corresponding rectangular image. Can be decided. In addition, when the position in the rectangular image corresponding to the pixel of the image in the original area is shifted from the pixel position of the rectangular image, the gray value of the pixel in the image in the original area corresponds to, for example, It can be determined by interpolation processing based on the gray value of the surrounding pixels at the position of the rectangular image.

  Furthermore, in the image processing method according to the present invention, an inverse conversion table representing a position in the rectangular image corresponding to each pixel of the image in the original area is created in advance, and the position inverse conversion step includes the inverse conversion step. A position in the rectangular image corresponding to each pixel of the image in the original area can be obtained using a table.

  With such a configuration, the position in the rectangular image corresponding to each pixel in the image in the original area is determined using the inverse conversion table, so the rectangular image corresponding to each pixel in the image in the original area The position at can be obtained at higher speed.

  Further, in the image processing method according to the present invention, the image acquisition step acquires an image including a surface image of a disk-shaped semiconductor wafer as the processed image, and the edge image specifying step includes an image of the semiconductor wafer surface image. An image of all or part of a ring-shaped region having a predetermined width in a direction orthogonal to a circular line along the edge is specified as an edge image, and the inverse transforming step includes the rectangle subjected to the one-dimensional image processing. The processed edge image can be generated by inversely transforming the image into an image of all or part of the original ring-shaped region.

  With such a configuration, in the image to be processed, the whole or part of the ring-shaped region including the circular line set along the edge of the semiconductor wafer surface image is substantially in the direction along the circular line. A processed edge image subjected to sequential image processing can be obtained.

  Furthermore, in the image processing method according to the present invention, in the edge image specifying step, the entire ring-shaped region having a first predetermined width inside the circular line and a second predetermined width outside the circular line. Alternatively, a part of the image can be specified as the edge image.

  With this configuration, even if a circular line that exactly matches the edge of the semiconductor wafer surface image is not set, all or part of the ring-shaped region that always includes the edge of the semiconductor wafer surface image is displayed as the edge image. Can be specified as

  Further, in the image processing method according to the present invention, the converting step may include converting the edge image into one of the two dividing lines parallel to the edge line of the object image of the edge image. And pixels arranged in a rectangle corresponding to each position set at a predetermined interval from each position to a normal line perpendicular to the lane marking line to the other lane marking line. It can be configured to convert to a rectangular image.

  With such a configuration, each of the pixels arranged in a rectangular shape obtains a rectangular image corresponding to the position in the edge image that is partitioned by two partition lines parallel to the edge line of the object image. be able to.

  Further, in the image processing method according to the present invention, the conversion step may include the step of converting the edge image into a longer partition of two partition lines parallel to the edge image of the object image across the edge line. Pixels arranged in a rectangle corresponding to each position set at a predetermined interval from each position set at a predetermined interval along the line and each pixel to a shorter division line on the normal line orthogonal to the division line It can comprise so that it may convert into the rectangular image which consists of.

  With such a configuration, each position in the edge image is set based on each position set at a predetermined interval along the longer lane line of the two lane lines that divide the edge image. More positions can be set in the edge image. As a result, it can be converted into a rectangular image composed of more pixels.

  Further, in the image processing method according to the present invention, one row of the rectangular image associated with each position along the partition line in the edge image and all positions set on the normal passing through the position. A conversion table for associating pixels in minutes, and the conversion step includes a step of determining a position in the edge image corresponding to each pixel of the rectangular image using the conversion table. can do.

  With such a configuration, the conversion table does not associate each of the positions in the edge image with the pixels of the rectangular image, but maps the position to each position along one division line in the edge image. Since the pixels corresponding to one column of the rectangular image associated with all the positions set on the normal line passing through are associated with each other, the conversion table can be configured on a smaller scale. It becomes possible to reduce the capacity of the memory for storing.

  An image processing method according to the present invention includes an image acquisition step of acquiring a processed image divided by two parallel dividing lines, and the processed image of the two parallel dividing lines of the processed image. A rectangular array corresponding to each position set at a predetermined interval along each lane line and each position set at a predetermined interval from each position to a normal line perpendicular to the lane line to the other lane line The conversion step of converting into a rectangular image made up of the processed pixels is provided.

  With such a configuration, it is possible to convert an image to be processed partitioned by two parallel partition lines into a rectangular image in which pixels are arranged in a rectangular shape.

  In the image processing method according to the present invention, each position along the partition line in the image to be processed corresponds to one column of the rectangular image associated with all positions set on the normal passing through the position. A conversion table for associating pixels is created in advance, and the conversion step includes a step of determining a position in the processed image corresponding to each pixel of the rectangular image using the conversion table. Can do.

  With such a configuration, the conversion table does not associate each of the positions in the edge image with the pixels of the rectangular image, but maps the position to each position along one division line in the edge image. Since the pixels corresponding to one column of the rectangular image associated with all the positions set on the normal line passing through are associated with each other, the conversion table can be configured on a smaller scale. It becomes possible to reduce the capacity of the memory for storing.

  An image processing apparatus according to the present invention includes a photographing unit that photographs a surface of an object, and a processing unit that processes a photographed image obtained by the photographing unit, and the processing unit is configured by grayscale values in pixel units. Image acquisition means for acquiring a processed image including the surface image of the object, and an edge line of the object surface image set in the processed image, and having a predetermined width in a direction orthogonal to the edge line Edge image specifying means for specifying an image of a region as an edge image, conversion means for converting the edge image into a rectangular image having a width corresponding to the predetermined width, and the rectangular image in the width direction The image processing means for performing one-dimensional image processing for sequentially processing the gray values of a plurality of pixels arranged in the orthogonal direction.

  With such a configuration, the processing unit includes the edge line of the region including the edge line of the object surface image set in the image to be processed and having a predetermined width in a direction orthogonal to the edge line, and a width corresponding to the predetermined width. The rectangular image is subjected to one-dimensional image processing for sequentially processing the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width of the rectangular image. In the one-dimensional image processing for the rectangular image, the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image are sequentially processed. Therefore, the rectangular image is a straight line corresponding to the edge line of the object surface image. Are sequentially processed in the direction along the line.

  In the image processing apparatus according to the present invention, the image processing apparatus further includes inverse conversion means for inversely converting the rectangular image subjected to the one-dimensional image processing into an image of an original region to generate a processed edge image. can do.

  With such a configuration, a processed edge image is generated by inversely converting the rectangular image subjected to the one-dimensional image processing into an image of the original region. As described above, since the one-dimensional image processing for the rectangular image is performed in the direction along the straight line corresponding to the edge line set in the object surface image, the one-dimensional image processed rectangular image is inversely transformed. The obtained processed edge image is substantially subjected to image processing along the edge line of the object surface image.

  In the image processing apparatus according to the present invention, the conversion unit includes a position conversion unit that acquires a position in the edge image corresponding to each pixel of the rectangular image, and a shade of each pixel in the edge image. Means for determining a gray value of a pixel of the rectangular image corresponding to each position of the acquired edge image based on the value.

  With such a configuration, the processing unit acquires the position in the edge image corresponding to each pixel constituting the rectangular image, and based on the gray value of each pixel in the edge image, the acquired edge portion The gray value of the pixel of the rectangular image corresponding to each position of the image is determined.

  Furthermore, in the image processing apparatus according to the present invention, the image processing apparatus further includes a storage unit that stores a conversion table representing a position in the edge image corresponding to each pixel of the rectangular image, and the position conversion unit stores the conversion table. It can be set as the structure which acquires the position in the said edge image corresponding to each pixel of the said rectangular image using.

  With such a configuration, the processing unit obtains the position in the edge image corresponding to each pixel of the rectangular image using the conversion table stored in the storage means, so that each pixel in the rectangular image in the processing unit. The position in the edge image corresponding to can be obtained at higher speed.

  Further, in the image processing apparatus according to the present invention, the photographing unit photographs a surface of a disk-shaped semiconductor wafer, the image obtaining unit obtains an image including the semiconductor wafer surface image as the processed image, The edge image specifying means specifies all or part of an image of a ring-shaped region having a predetermined width in a direction orthogonal to a circular line along the edge of the semiconductor wafer surface image as the edge image, and the inverse conversion means Can be configured to inversely convert the rectangular image subjected to the one-dimensional image processing into an image of all or part of the original ring-shaped region to generate a processed edge image.

  With such a configuration, the processing unit substantially follows the circular line with respect to all or a part of the image of the ring-shaped region including the circular line set along the edge of the surface image of the disk-shaped semiconductor wafer. It is possible to obtain a processed edge image that has been sequentially subjected to image processing in the selected direction.

  Further, in the image processing apparatus according to the present invention, the edge image specifying means has a ring-shaped region that has a first predetermined width inside the circular line and a second predetermined width outside the circular line. Alternatively, a part of the image can be specified as the edge image.

  With such a configuration, even if a circular line that exactly matches the edge of the semiconductor wafer surface image is not set, the processing unit can image all or part of the ring-shaped region that always includes the edge of the semiconductor wafer surface image. Can be specified as the edge image.

  An image processing apparatus according to the present invention includes a photographing unit that photographs a surface of an object, and a processing unit that processes a photographed image obtained by the photographing unit, and the processing unit includes two parallel parts from the photographed image. Image acquisition means for acquiring an image to be processed divided by one lane line, and the image to be processed at a predetermined interval along one of the two parallel lane lines of the image to be processed. Conversion means for converting into a rectangular image composed of pixels arranged in a rectangular shape corresponding to each position set at a predetermined interval from each position set to a normal line perpendicular to the lane marking line to the other lane marking line. It becomes the composition which has.

  In the image processing apparatus according to the present invention, each position of the rectangular image corresponding to all positions set on the normal passing through the position at each position along the partition line in the processed image. A storage unit that stores a conversion table that associates pixels; and the conversion unit includes a unit that determines a position in the image to be processed corresponding to each pixel of the rectangular image using the conversion table. can do.

  With such a configuration, with such a configuration, the conversion table does not associate each position in the edge image with each pixel of the rectangular image, but each one along one partition line in the edge image. Since the pixels corresponding to one column of the rectangular image associated with all the positions set on the normal line passing through the position are associated with the position, the conversion table can be configured on a smaller scale. As a result, the capacity of the storage means for storing it can be reduced.

  The surface inspection apparatus according to the present invention includes an imaging unit that images the surface of an object, and a processing unit that processes a captured image obtained by the imaging unit and detects a defect on the object surface. The unit includes an image acquisition unit configured to obtain a processed image including a surface image of the object, and an edge line of the object surface image set in the processed image. Edge image specifying means for specifying an image of a region having a predetermined width in a direction orthogonal to the line as an edge image, conversion means for converting the edge image into a rectangular image having a width corresponding to the predetermined width, and First image processing means for performing one-dimensional image processing for sequentially processing grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image; and the rectangular image subjected to the one-dimensional image processing. Reverse to original area image Inverse conversion means for generating a processed edge image in turn, and a body part image obtained by removing the edge image from the object surface image, sequentially processing the gray values of a plurality of pixels arranged two-dimensionally 2 A second image processing unit that generates a processed main body image by performing dimensional image processing; and a defect specifying unit that specifies a defective portion in the processed edge image and the processed product main body image. .

  With such a configuration, the processing unit includes the edge line of the region including the edge line of the object surface image set in the image to be processed and having a predetermined width in a direction orthogonal to the edge line, and a width corresponding to the predetermined width. The rectangular image is subjected to one-dimensional image processing for sequentially processing the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width of the rectangular image. In the one-dimensional image processing for the rectangular image, the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image are sequentially processed. Therefore, the rectangular image is a straight line corresponding to the edge line of the object surface image. Are sequentially processed in the direction along the line.

  Then, the processing unit generates a processed edge image by inversely converting the rectangular image subjected to the one-dimensional image processing into an image of the original region. As described above, since the one-dimensional image processing for the rectangular image is processing in a direction along the straight line in the rectangular image corresponding to the edge line of the object surface image, the one-dimensional image processed rectangular image is The processed edge image obtained by the inverse transformation is substantially sequentially processed in the direction along the edge line of the object surface image.

  In addition, the processing unit performs a two-dimensional image processing for sequentially processing the grayscale values of a plurality of pixels arranged two-dimensionally on the main body image obtained by removing the edge image from the object surface image, thereby processing the processed main body image. And a defective portion is specified in the processed edge image and the processed main body image.

  As described above, both the edge line image and the main body image of the object surface image are identified in the image processed state, so that the inspection of the defect over the entire object surface can be performed with higher accuracy. it can.

  15. The surface inspection apparatus according to claim 14, wherein the imaging unit images a surface of a disk-shaped semiconductor wafer, and the processing unit detects a defect on the surface of the semiconductor wafer. The image acquisition means acquires an image including the surface image of the semiconductor wafer as the processed image, and the edge image specifying means is a direction orthogonal to a circular line along the edge of the semiconductor wafer surface image. The whole or part of the ring-shaped region having a predetermined width is specified as an edge image, and the inverse transforming means converts the rectangular image subjected to the one-dimensional image processing to all or the original ring-shaped region. The second image processing means generates the processed edge image by inversely converting into a part of the image, and the second image processing means applies the secondary image to the main body image obtained by removing the edge image from the semiconductor wafer surface image. Configured to generate a processed body section image is subjected to image processing.

  With such a configuration, all or a part of the ring-shaped region including the circular line set along the edge of the disk-shaped semiconductor wafer surface image is sequentially subjected to image processing in the direction along the circular line. In both the processed edge image and the processed table main body image obtained by performing the two-dimensional image processing on the main body image obtained by removing the edge image from the semiconductor wafer surface image, the defect portion is specified. As a result, it is possible to inspect defects over the entire surface of the semiconductor wafer with higher accuracy.

  According to the image processing method and the image processing apparatus according to the present invention, the edge image including the edge line of the object surface image specified in the image to be processed is converted into a rectangular image, and one-dimensional image processing is performed on the rectangular image. Is given. In the one-dimensional image processing for the rectangular image, the gray values of a plurality of pixels arranged in the direction orthogonal to the width direction of the rectangular image are sequentially processed. This corresponds to an image obtained by sequentially performing image processing in the direction along the edge line. Then, if the processed rectangular image is inversely transformed, it is possible to easily obtain an image that has undergone appropriate image processing in the region of the edge portion of the object surface image.

It is a side view which shows the structure of the mechanism part in the surface inspection apparatus of a semiconductor wafer. It is a front view which shows the structure of the mechanism part in the surface inspection apparatus of a semiconductor wafer. It is a figure which shows the state of the scanning of the semiconductor wafer surface by a camera unit (CCD line sensor). It is a block diagram which shows the control system in the surface inspection apparatus of a semiconductor wafer. It is a flowchart which shows the procedure of a registration process. It is a figure which shows an example of a picked-up image. It is a figure which shows an example of the picked-up image by which position correction was carried out. It is a figure which shows the state in which the circular line was set to the edge of the wafer surface image in the picked-up image. It is a figure which shows the ring-shaped area | region containing the edge line of the wafer surface image set in the picked-up image. It is a figure which shows the pixel structure of the rectangular image in which the image (edge part image) of a ring-shaped area | region should be converted. It is a figure which shows the relationship between the edge part image specified in the picked-up image, and the rectangular image obtained by converting it. It is a figure which shows the edge part image of the state by which the normal line NL was set to arbitrary angle positions (theta). It is the figure which expanded and showed angle range 0 degree-90 degrees of FIG. 10A. It is a figure which expands and shows the position set to the outer periphery of an edge part image. It is a figure which shows the position set on the normal line NL. It is a figure which expands and shows the position set on the normal line NL. It is a figure which shows the relationship between each pixel of a rectangular image, and the position of an edge image. It is a figure which shows an example of the information content which a conversion table has. It is a figure which shows the relationship between a rectangular image and the image (edge part image) of the ring-shaped area | region obtained by reversely transforming it. It is a flowchart which shows the procedure of an inspection process. It is a figure which shows an example of the picked-up image of the semiconductor wafer used as a test object. It is a figure which shows an example of the picked-up image by which position correction was carried out. It is a figure which shows the relationship between the edge image specified by the picked-up image, and the rectangular image obtained by converting it. It is a figure which shows the rectangular image before a one-dimensional image process, and the rectangular image after a one-dimensional image process. It is a figure which shows the relationship between the processed rectangular image and the processed edge image obtained by reversely transforming it. It is a figure which shows an example of the wafer surface image in which the detected defect was represented.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  An image processing apparatus that performs image processing according to the image processing method according to the present invention is configured, for example, in a semiconductor wafer surface inspection apparatus. The mechanism part of this surface inspection apparatus is configured as shown in FIGS. FIG. 1 is a side view showing the configuration of the mechanism, and FIG. 2 is a front view thereof.

  1 and 2, the surface inspection apparatus includes a base 50, and a carriage 60 and a shift movement mechanism 70 are provided in the base 50 (see particularly FIG. 1). The shift moving mechanism 70 includes a stage 71, and the carriage 60 can reciprocate by moving on the stage 71 (moving left and right in FIG. 1 and perpendicular to the paper surface in FIG. 2). The shift movement mechanism 70 also has a drive unit 72 including a motor, a gear mechanism, and the like. The drive unit 72 causes the stage 71 to move in a direction perpendicular to the movement direction of the carriage 60 due to self-running (the left-right direction in FIG. 1 in a direction perpendicular to the paper surface).

  The carriage 60 is provided with a support mechanism 20 that supports a semiconductor wafer (hereinafter simply referred to as a wafer) 10 to be inspected, and the support mechanism 20 protrudes from the upper surface of the base 50. The support mechanism 20 includes a circular table 21 on which the wafer 10 is set, and two support legs 22 a and 22 b that support the table 21 side by side along a direction orthogonal to the moving direction of the carriage 60. These support legs 22 a and 22 b are fixed to the carriage 60. Although not shown, a guide mechanism is provided on the upper surface of the base 50 to guide the support legs 22a and 22b in the moving direction when the carriage 60 is self-propelled. When the carriage 60 shifts in the direction orthogonal to the self-running direction by 70, a shift guide mechanism for guiding the support legs 22a and 22b in that direction is provided.

  An arch-shaped frame 55 is provided at a substantially central portion in the self-running direction of the carriage 60 on the upper surface of the base 50. A camera unit 30 is provided in a substantially central portion of the frame 55 facing downward. An illumination unit 31 is provided in the vicinity of the installation position of the camera unit 30 on the frame 55. The illumination unit 31 illuminates the surface of the wafer 10 supported by the support mechanism 20 that moves below the frame 55.

  The camera unit 30 has a line sensor (for example, a one-dimensional CCD image sensor), and photographs the surface of the wafer 10 supported by the support mechanism 20 that moves as the carriage 60 moves. Specifically, as shown in FIG. 3, the camera unit 30 images the entire surface of the wafer 10 by four scans.

  In FIG. 3, first, when the carriage 60 moves in the direction A1 at the shift initial position Ps1, the camera unit 30 scans a predetermined width (corresponding to the length of the line sensor) on the surface of the wafer 10 in the direction S1. When the scanning at the shift initial position Ps1 is completed, the carriage 60 is shifted in the direction B to the shift position Ps2 by the shift moving mechanism 70. Next, at the shift position Ps2, when the carriage 60 moves in the direction A2 opposite to the direction A1, the camera unit 30 scans the wafer 10 in the direction S2 (reverse to the direction S1). When the scanning at the shift position Ps2 is completed, the carriage 60 is further shifted in the direction B to the shift position Ps3 by the shift mechanism 70. When the carriage 60 moves in the direction A1 at the shift position Ps3, the camera unit 30 scans the wafer 10 in the direction S1. When the scanning at the shift position Ps3 is completed, the carriage is further shifted to the shift position Ps4 by the shift mechanism 70. At this shift position Ps4, when the carriage 60 moves in the direction A2, the camera unit 30 scans in the wafer S2 direction. As described above, while the carriage 60 reciprocates in the directions A1 and A2 due to self-running, the camera unit 30 scans the surface of the wafer 10 four times while repeating the shift movement in the direction B by the shift movement mechanism 70. To do. In the process, the camera unit 30 images the surface of the wafer 10 and sequentially outputs image signals.

  The control system of the surface inspection apparatus is configured as shown in FIG.

  In FIG. 4, the processing unit 100 inputs the image signal from the camera unit 30 described above, and processes the image signal as information representing the surface image of the wafer 10. Further, a drive control unit 120 is connected to the processing unit 100, and the drive control unit 120 performs drive control of the moving mechanism 200 including the carriage 60 and the shift moving mechanism 70 described above under the control of the processing unit 100. . Further, an operation unit 111 and a display unit 112 are connected to the processing unit 100. The processing unit 100 that inputs an operation signal from the operation unit 111 operated by the user executes various processes based on the operation signal. The processing unit 100 causes the display unit 112 to display various types of information, such as displaying a captured image including the surface image of the wafer 10 on the display unit 112 based on the image data obtained by the processing.

  Next, processing for detecting defects on the surface of the wafer 10 will be described.

  First, processing (registration processing) related to registration of various information related to image processing of the surface image of the wafer 10 is performed.

  FIG. 5 shows a processing unit 100 in which the surface of the wafer 10 representing the wafer to be inspected is photographed by the camera unit 30 in the above-described mechanism (see FIGS. 1 to 3) and an image signal from the camera unit 30 is input. The process is executed according to the procedure shown.

  In FIG. 5, the processing unit 100 converts the image signal sequentially supplied from the camera unit 30 into image data composed of grayscale values in pixel units, and includes a wafer surface image (data) representing the entire surface of the wafer 10. A captured image is acquired as an image to be processed (S1). The captured image including the wafer surface image is developed in an image memory (not shown). For example, as shown in FIG. 6A, when the processing unit 100 acquires a captured image (processing target image) I including a wafer surface image I10, the captured image I is displayed on the image memory as X (lateral coordinate axis), Y The position is corrected based on the respective references (vertical coordinate axis) and θ (counterclockwise angle) (S2). For example, as shown in FIG. 6B, the position is corrected so that the notch portion I11 of the wafer surface image I10 is positioned in a predetermined direction. The processing unit 100 causes the display unit 112 to display the captured image I including the wafer surface image I10 whose position has been corrected.

For the captured image I displayed on the display unit 112, the user operates the operation unit 111, for example, as shown in FIG. 6C, a circular line C0 (edge line: along the edge Edg of the wafer surface image I10). Set up a chain line). When a predetermined determination operation is performed in the operation unit 111 in a state where the circular line C0 is set, the processing unit 100 determines the set radius R of the circular line C0 and the position (Xo, Yo) of the center O thereof. Obtain (S3). Next, as shown in FIG. 7, the processing unit 100 is concentric with the circular line C0 specified by the radius R and the position (Xo, Yo) of the center O, and inside the radius (R−L1). A circular division line C1 and an outer circular division line C2 having a radius (R + L2) are set concentrically with the circular line C0. As a result, the processing unit 100 is partitioned by the inner circular partition line C1 and the outer circular partition line C2 that are parallel to the circular line C0 and has a width from the circular line C0 to the inside of the wafer surface image I10. A ring-shaped region E _Ling having a width L2 outside of L1 is recognized, and an image in the ring-shaped region E _Ling in the captured image I is specified as the edge image IE (S4). Note that L1 and L2 are set, for example, by the operation of the operation unit 11 by the user.

Next, the processing unit 100 creates a conversion table (HLUT) to be used for converting the edge image IE of the ring-shaped region E _Ling into a rectangular image IR having a width corresponding to the width (L1 + L2) ( S5). This conversion table (HLUT) is a ring-shaped edge image having an outer circumference 2π (R + L2) (the length of the outer circular division line C2) and an inner circumference 2π (R−L1) (the length of the inner circular division line C1). Each position of IE is associated with a pixel Px of a rectangular image IR having a length and width corresponding to the outer periphery 2π (R + L2) and width (L1 + L2) of the edge image IE as shown in FIG. Specifically, as shown in FIG. 9, each position of the first edge image portion IE (1) in the first angle range (0 ° ≦ θ <90 °) in the ring-shaped edge image IE is long. Is associated with a pixel of the first rectangular image portion IR (1) having a length corresponding to the first angle range (0 ° ≦ θ <90 °) of 2π (R + L2), and Similarly, each position of the second edge image portion IE (2) in the two angle range (90 ° ≦ θ <180 °) has a length corresponding to the second angle range (90 ° ≦ θ <180 °). Corresponding to the pixel of the second rectangular image portion IR (2). Further, each position of the third edge image portion IE (3) in the third angle range (180 ° ≦ θ <270 °) in the ring-shaped edge image IE corresponds to the position of the length 2π (R + L2). Corresponding to the pixels of the third rectangular image portion IR (3) having a length corresponding to the third angle range (180 ° ≦ θ <270 °), and further, the fourth angle range (270 ° ≦ 270 ° in the edge image IE). Similarly, each position of the fourth edge image portion IE (4) of θ <360 °, 0 ° has a length corresponding to the fourth angle range (270 ° ≦ θ <360 °, 0 °). Corresponding to the pixels of the four rectangular image portion IR (4).

  The conversion table represents the position of the edge image IE corresponding to each pixel of the rectangular image IR, and is created as follows.

In the ring-shaped edge image IE having an outer peripheral edge (outer circular division line C2) and an inner peripheral edge (inner circular division line C1) as shown in FIG. 10A and FIG. 10B, as shown in FIG. N positions P ₀ (X ₀ , Y ₀ ), P ₁ (X ₁ , Y ₁ ) at equal intervals δ (unit length based on the resolution of one pixel) along the outer circular division line C2),. P _i (X _i , Y _i ),..., P _n-1 (X _n-1 , Y _n-1 ) is determined. Note that n = 2π (R + L2) / δ. Next, as shown in FIG. 10A, FIG. 10B, and FIG. 11B, it passes through each position P _j set on the outer peripheral edge (outer circular division line C2), and the outer peripheral edge (outer circular division line C2) and inner peripheral edge ( When a straight line (hereinafter referred to as a normal line) NL (θ) orthogonal to the inner circular partition line C1) is set, the intersection A with the outer peripheral edge (C2) on the normal line NL (θ) and the inner peripheral edge _M positions P _{j1Δ to} P _jmΔ can be set at equal intervals (δ) between the intersection B with (C1). Note that m = (L1 + L2) / δ. Specifically, as shown in FIG. 12, on the normal NL (θ), ΔX _j (= δ · cos θ) sequentially from the position P _{j of the} point A in the X direction and ΔY _j ( Each position P _{j1Δ to} P _jmΔ shifted by = δ · sin θ) can be set until the point B is reached. The coordinate values of each position are P _j (X _j , Y _j ), P _j1Δ (X _j −ΔX _j , Y _j −ΔY _j ),..., P _jmΔ (X _j −mΔX _j , Y _j −mΔY _j ).

As described above, each position set in the edge image IE is associated with a pixel of the rectangular image IR shown in FIG. Specifically, as shown in FIG. 13, each of the n pixels Px ₀ to Px ₀ aligned along the side V (C2) of the rectangular image IR corresponding to the outer peripheral edge (outer circular division line C2) of the edge image IE. Positions P ₀ (X ₀ , Y ₀ ) to P _n-1 (X _n-1 , Y _n- ) where Px _n-1 is set along the outer peripheral edge (outer circular division line C2) of the edge image IE. ₁ ), n pixels Px _{01 to} Px _{(n−1) 1 that} are aligned by one pixel from the side V (C2) of the rectangular image IR are the outer peripheral edge (outer circular shape ₎ of the edge image IE. Positions P _01Δ (X ₀ −ΔX ₀ , Y ₀ −ΔY ₀ ) to P _{(n−1) 1Δ} (X _n−1 −ΔX _n−1 , Y _{n )} set inward by the interval δ from the lane _marking C 2) ₋₁₋ ΔY _n−1 ), and _n similarly arranged along the side V (C1) of the rectangular image IR corresponding to the inner peripheral edge (inner circular division line C1) of the edge image IE. number of pixels _{Px 0m ~Px (n-1)} m Inner peripheral edge position P _0Emuderuta set along the (inner circular partition line C1) of the edge image _{IE (X 0 -mΔX 0, Y} 0 -mΔY 0) ~P (n-1) mΔ (X n-1 −mΔX _n−1 , Y _n−1 −mΔY _n−1 ).

In consideration of the above-described relationship, the conversion table (HLUT) includes pixels (Px _{0 to} Px _0m ), (Px _{1 to} Px _1m ), (Px _{1 to} Px _1m ), one column in the width direction in the rectangular image IR, as shown in FIG. ..., (Px _{n-1 to} Px _{(n-1) m} ), positions P _{0 to} P _n- set along the outer peripheral edge (outer circular division line C2) of the edge image IE. numerical containing _one coordinate values _{(X 0, Y 0, ΔX} 0, ΔY 0), (X 1, Y 1, ΔX 1, ΔY 1), ···, (X n-1, Y n-1, ΔX _n−1 , ΔY _n−1 ) are associated with each other. By using such a conversion table (HLUT), each position of the edge image IE corresponding to each pixel (Px _{0 to} Px _om ) in the first column aligned in the width direction in the rectangular image IR becomes (X ₀ , (Y ₀ ), (X ₀ −ΔX ₀ , Y ₀ −ΔY ₀ ),..., (X ₀ −mΔX ₀ , Y ₀ −mΔY ₀ ). Each position of the edge image IE corresponding to each pixel (Px _{i to} Px _im ) in the (i + 1) th column is (X _i , Y _i ), (X _i −ΔX _i , Y _i −ΔY _i ), .., (X _i −mΔX _i , Y _i −mΔY _i )

The conversion table (HLUT) is not the correspondence between all the positions set in the edge image IE and each pixel of the rectangular image IR as shown in FIG. 13, but the width direction of the rectangular image IR as shown in FIG. Positions P ₀ (X ₀ , Y ₀ ) to P _n-1 (X _n-1 , Y _n-1 ) along the outer edge line (outer circular division line C2) in the edge image IE. ) And the position interval Δ set on the normal line passing through each position, the conversion table (HLUT) can be configured on a smaller scale, and as a result, it is stored. The memory capacity can be reduced.

Returning to FIG. 5, when a conversion table (HLUT) representing the position of the edge image IE corresponding to each pixel of the rectangular image IR is generated in the relationship as described above, the processing unit 100 conversely An inverse conversion table (NLUT) representing the position of the rectangular image IR corresponding to each pixel of the partial image IE is generated (S6). As shown in FIG. 8, each position of the rectangular image IR having a length and width corresponding to the length (L1 + L2) and the width (L1 + L2) is represented by an element of the outer circumference 2π (R + L2) and the inner circumference 2π (R−L1). Are associated with the pixels of the image (edge image IE) of the ring-shaped region E _Ling . Specifically, as shown in FIG. 15, the first rectangular image portion IR (1) having a length corresponding to the first angle range (0 ° ≦ θ <90 °) in the length 2π (R + L2). Each position is associated with each pixel of the first edge image portion IE (1) in the first angle range (0 ° ≦ θ <90 °) in the ring-shaped edge image IE, and similarly the second angle range. Each position of the second rectangular image portion IR (2) having a length corresponding to (90 ° ≦ θ <180 °) is represented by the second angle range (90 ° ≦ θ <180 °) in the edge image IE. Corresponding to the pixels of the two edge image portion IE (2). The inverse conversion table also indicates each position of the third rectangular image portion IR (3) having a length corresponding to the third angle range (180 ° ≦ θ <270 °) of the length 2π (R + L2). Corresponding to the pixels of the third edge image portion IE (3) in the third angle range (180 ° ≦ θ <270 °) in the ring-shaped edge image IE, and further, the fourth angle range (270 ° ≦ θ Each position of the fourth rectangular image portion IR (4) having a length corresponding to <360 °, 0 °) is set in the fourth angle range (270 ° ≦ θ <360 °, 0 °) in the edge image IE. Corresponding to the pixels of the fourth edge image portion IE (4).

  The method for generating the inverse conversion table (NLUT) is the same as that of the conversion table (HLUT) described above.

When the conversion table (HLUT) and the inverse conversion table (NLUT) are generated, the processing unit 100 uses the conversion table (HLUT) to specify the edge specified in the captured image I including the wafer surface image I10. The image IE is converted into a rectangular image IR, and the rectangular image IR obtained by the conversion using the inverse conversion table (NLUT) is further converted back into the original ring-shaped area E _Ling image (edge image IE). (S7). A captured image I including an image obtained by the inverse transformation is displayed on the display unit 112. If there is no problem in the image portion of the captured image I displayed on the display unit 112, particularly in the ring-shaped region E _Ling , the user performs a predetermined registration execution operation with the operation unit 111. When the processing unit 100 receives an operation signal based on the registration execution operation from the operation unit 111, the processing unit 100 specifies the circular line C0 (edge line) along the edge of the wafer surface image I10 set as described above. Center position O (Xo, Yo) and radius (R), boundary line of ring-shaped region E _Ling (C1,
) From the circular line C0 representing the center line (Xo, Yo) obtained by removing the ring-shaped region E _Ling from the circular region of the conversion table (HLUT), inverse conversion table (NLUT), and wafer surface image I10. The circular main body region having the radius (R-L1) is registered in the internal memory (S8), and the process is terminated. This registration process can be performed for each diameter of the wafer to be inspected.

If there is any problem in the image portion of the captured image I obtained through the conversion and inverse conversion, particularly in the ring-shaped region E _Ling , the user again captures the surface of the wafer 10 and takes the wafer. A circular line C0 and widths L1 and L2 for the surface image I10 are determined, and a conversion table (HLUT) and an inverse conversion table (NLUT) can be generated.

  After the registration process is completed as described above, the surface inspection apparatus can perform the surface inspection of each wafer 10. The surface of the wafer 10 to be inspected is photographed by the camera unit 30 in the above-described mechanism (see FIGS. 1 to 3), and the processing unit 100 that inputs an image signal from the camera unit 30 performs processing according to the procedure shown in FIG. Execute.

  In FIG. 16, the processing unit 100 converts image signals sequentially supplied from the camera unit 30 that images the surface of the wafer 10 to be inspected into image data composed of grayscale values in pixel units, and the surface of the wafer 10. A photographed image including a wafer surface image (data) representing the whole is acquired as an image to be processed (S11). The image to be processed including the wafer surface image is developed in an image memory (not shown). For example, when the processing unit 100 acquires a captured image I including a wafer surface image I20 as shown in FIG. 17A, the captured image I is displayed on the image memory as X (horizontal coordinate axis), Y (vertical coordinate axis), and θ. The position is corrected based on each reference (counterclockwise angle) (S12). For example, as shown in FIG. 17B, the notch portion I21 of the wafer surface image I20 is positioned in a predetermined direction, and the wafer surface image I20 is positioned in the same manner as that during registration processing (see FIGS. 5 and 6B). The position of the captured image I after position correction is developed on the image memory. In this case, there is a defective portion D1 in the outer edge portion of the wafer surface image I20.

The processing unit 100 is position-corrected based on the center position (Xo, Yo) and radius (R) that specify the registered circular line C0 and the widths L1 and L2 that specify the ring-shaped region E _Ling . The inspection area is mapped in the wafer surface image I20 (S13). By this mapping, an image of the ring-shaped region E _Ling is specified as the edge image IE, and a circular main body region (radius) obtained by removing the ring-shaped region E _Ling (edge image IE) from the region of the wafer surface image I20. (R-L1) wafer main body image) is identified.

Next, the processing unit 100 converts the edge image IE, which is an image of the ring-shaped region E _Ling obtained by the mapping, into a rectangular image IR using the conversion table (HLUT) (see FIGS. 13 and 14) described above. (S14). In this conversion process, first, the position in the ring-shaped edge image IE corresponding to each pixel (see FIG. 8) of the rectangular image IR is acquired from the conversion table (HLUT). Next, the gray value of the position in the acquired edge image IE is acquired, and the gray value is determined as the gray value of the pixel of the rectangular image IR corresponding to the position. When the position in the edge image IE matches the pixel position in the edge image IE, the gray value of the pixel is determined as the gray value of the pixel in the corresponding rectangular image IR. Further, when the position in the edge image IE does not match the pixel position of the edge image IE, the gray value of the position is determined by interpolation processing using the gray value of the peripheral pixels at the position, and the gray value corresponds to the position value. It is determined as the gray value of the pixel of the rectangular image IR.

  In this way, the gray value of each pixel in the rectangular image IR is determined as the gray value of the position in the edge image IE associated with the conversion table (HLUT), so that the edge image IE becomes the rectangular image IR. Is converted to By this conversion processing, as shown in FIG. 18, the first edge image portion IE (1) in the first angle range (0 ° ≦ θ <90 °) in the edge image IE has a length of 2π (R + L2). The first rectangular image portion IR (1) having a length corresponding to the first angular range (0 ° ≦ θ <90 °) is converted into the second angular range (90 ° ≦ θ <in the edge image IE). 180 °) second edge image portion IE (2) is similarly converted into a second rectangular image portion IR (2) having a length corresponding to the second angle range (90 ° ≦ θ <180 °). The Further, the third edge image portion IE (3) in the third angle range (180 ° ≦ θ <270 °) in the ring-shaped edge image IE is the third angle of the length 2π (R + L2). It is converted into a third rectangular image portion IR (3) having a length corresponding to the range (180 ° ≦ θ <270 °), and further, the fourth angular range (270 ° ≦ θ <360 °, 0 in the edge image IE). The fourth edge image portion IE (4) at the same angle is the fourth rectangular image portion IR (4) having a length corresponding to the fourth angle range (270 ° ≦ θ <360 °, 0 °). Converted. In this case, the defect portion D1 present in the second edge image portion IE (2) is converted into the defect portion DR1 of the corresponding second rectangular image portion IR (2).

Returning to FIG. 16, when the processing unit 100 generates the rectangular image IR, the processing unit 100 performs one-dimensional filter processing on the rectangular image IR in order to emphasize the defective portion (S15). In this one-dimensional filter processing, the gray value of each target pixel corresponds to the direction orthogonal to the width direction of the rectangular image IR with respect to the target pixel, that is, the circular line C0 set as the edge line of the edge image IE. It is determined based on the gray value of the pixels lined up and down in the direction parallel to the horizontal line V (C0) (see FIG. 8). Such one-dimensional filtering process, the processed rectangular image IR _P in a state in which the defect portion is emphasized is obtained. Specifically, as shown in FIG. 19, the first processed rectangular image corresponding to the same angular range from the first rectangular image portion IR (1) corresponding to the first angular range (0 ° ≦ θ <90 °). IR _P (1) is generated, and from the second rectangular image portion IR (2) corresponding to the second angular range (90 ° ≦ θ <180 °), the second processed rectangular image IR _P ( 2) is generated, and the third processed rectangular image IR _P (3) corresponding to the same angular range is obtained from the third rectangular image portion IR (3) corresponding to the third angular range (180 ° ≦ θ <270 °). Further, a fourth processed rectangular image IR _P (4) corresponding to the same angular range is generated from the fourth rectangular image portion IR (4) corresponding to the fourth angular range (270 ° ≦ θ <360 °, 0 °). ) Is generated. In this case, the defective portion DR1 included in the second rectangular image portion IR (2) is emphasized and generated as the defective portion DR1 _P in the second processed rectangular image portion IR _P (2).

When the rectangular image IR is subjected to the one-dimensional filter processing to obtain the processed rectangular image IR _P , the processing unit 100 uses the inverse conversion table (NLUT) to convert the processed rectangular image IR _P to the original ring shape. Inverse transformation to the image of the region E _Ling (S16). In the process of inverse transformation, first, the position of the processed rectangular image IR _P corresponding to each pixel of the original ring-shaped region E _Ling is obtained from the inverse conversion table (NLUT). Then, the acquired grayscale values of the position in the acquired the processed rectangular image IR _P, the gray value is determined as the gray value of the pixel in the ring region E _Ling corresponding to the position. When a position in the processed rectangular image IR _P matches the pixel position in the processed rectangular image IR _P, the gray value of the pixel is determined as the gray value of the pixels of the corresponding ring-shaped region E _Ling. Further, when the position of the processed rectangular image IR _P does not match the pixel position in the processed rectangular image IR _P, gray value of the position is determined by an interpolation process according to the gray value of the peripheral pixel of the position, the shade The value is determined as the gray value of the pixel in the corresponding ring-shaped region E _Ling .

In this way, by being determined to gray value of the position of the processed rectangular image IR _P associated with gray value of each pixel of the ring-shaped region E _Ling inverse conversion table (NLUT), the processed rectangular image The IR _P is converted into an image of the ring-shaped region E _Ling , and a processed edge image IE _P is obtained. As a result of this inverse transformation processing, as shown in FIG. 20, the first processed rectangular image portion IR _P (1) corresponding to the first angular range (0 ° ≦ θ <90 °) is converted into the first processing within the same angular range. The second processed rectangular image part IR _P (2) corresponding to the second angle range (90 ° ≦ θ <180 °) is inversely transformed into the finished edge image part IE _P (1), 2 The processed edge image portion IE _P (2) is converted into a third processed rectangular image portion IR _P (3) corresponding to the third angle range (180 ° ≦ θ <270 °). The fourth processed rectangular image portion IR _P (4) converted to the third processed edge image portion IE _P (3) and further corresponding to the fourth angle range (270 ° ≦ θ <360 °, 0 °). Is converted into the fourth processed edge image portion IE _P (4) in the same angle range. In this case, the defect portion DR1 _P highlighted in the second processed rectangular image portion IR _P (2) is converted into the defect portion D1 _P emphasized in the second processed edge image IE _P (2). Is done.

Returning to FIG. 16, the processing unit 100 removes the ring region E _Ling from the region of the wafer surface image I20 obtained by the mapping (see S13) following the generation of the processed edge image IE _P described above. Two-dimensional filter processing is performed on the image of the circular main body region (S17). Specifically, in the image of the circular main body region (wafer main body image), the gray value of each pixel of interest is the gray value of its surrounding pixels (for example, eight pixels surrounding the pixel of interest) arranged two-dimensionally. It is decided based on. By performing such a two-dimensional filter process, a processed wafer main body image is obtained. In the processed wafer main body image, the defective portion is emphasized. Next, the processing unit 100 synthesizes the processed edge image IE _P and the processed wafer main body image to generate a single processed wafer surface image I20 _P (S18). Then, the processing unit 100, the defect detection process is performed on the generated processed wafer surface images I20 _P (S19). For example, the defect portion is specified based on the density distribution of the processed wafer surface images I20 _P. The processing unit 100 is information for specifying the defect part obtained by the defect detection process, that is, defect information (for example, defect number (NO.), Center coordinates of the defect part, area of the defect part, flatness of the defect part). , major and minor axes of the ellipse approximating the defect, to produce a defect classification number (NO.) or the like), and stores the defect information with the processed wafer surface images I20 _P (S20). Then, as shown in FIG. 21, the processing unit 100 causes the display unit 112 to display the defective portions D1 _P , D2 _P , D3 _P , D4 _P specified by the defect information together with the processed wafer surface image I20 _P. . In FIG. 21, the defective portion D1 _P is specified on the processed edge image IE _P , and the other defective portions D2 _P , D3 _P and D4 _P are the processed wafer main body image. As specified above.

In the image processing performed by the wafer surface inspection apparatus described above, the edge image IE including the circular line C0 along the edge of the wafer surface image I10 specified in the photographed image I (image to be processed) is converted into the rectangular image IR. After the conversion, the one-dimensional filter process is performed on the rectangular image IR. Since this one-dimensional filter process is a process in the direction along the straight line V (Co) corresponding to the circular line Co set with respect to the edge of the wafer surface image I20, the process subjected to the one-dimensional filter process processing Sumien unit image IE _P obtained already rectangular image IR _P is inverse transform becomes the filter along the edge lines of substantially the wafer surface image I20 is applied. Accordingly, it is possible to easily obtain an image (processed edge image IE _P ) that has been subjected to appropriate filtering in the region of the edge portion of the wafer surface image I20.

In addition, since an image (processed edge image IE _P ) that has been appropriately filtered is obtained in the region of the edge portion of the wafer surface image I20, the defect portion D1 _p in the edge portion of the wafer surface image I20 is also obtained. Like the defective portions D2 _P , D3 _P , and D4 _{P in} the inner wafer main body image, it can be accurately detected.

In the above-described embodiment of the present invention, the image in the ring-shaped region E _Ling including the circular line C0 as the edge line set in the wafer surface image I20 is the edge image IE. It is also possible to process a part of the _Ling image as the edge image IE. Even if the edge line of the object surface is not a circle but an arbitrary curve, an edge line along the edge curve is set, and an image of an area having a predetermined width in the direction orthogonal to the edge line is It can also be specified as a partial image. In this case, the edge image is converted into a rectangular image on the basis of the correspondence between each position of the edge image in the arbitrary curved shape and each pixel of the rectangular image, and the processed rectangular image is converted into the original curved shape. Is converted back to an image of the region (processed edge image).

  In the image processing method and the image processing apparatus according to the present invention, the one-dimensional image processing performed on the rectangular image IR is not limited to the filter processing, and a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image IR. There is no particular limitation as long as it is a one-dimensional image process that sequentially processes the gray values.

In the embodiment of the present invention described above, after obtaining the processed rectangular image IR _P by applying 1-dimensional image processing (1D filtering) for the converted rectangular image IR from the edge image IE, further, generating the processed rectangular image IR _P inverse transform to the processed edge image IE _P, it detects the defect in the processed edge image IE _P (actually a composite image). However, simply in such a case to detect only particular the presence or absence of a defect in the edge image IE of the processed image, without the need to inverse transform the processed rectangular image IR _P, the processing of the defect detected in the processed rectangular image IR _P It can be performed.

  Furthermore, in the above-described embodiment of the present invention, the captured image of the semiconductor wafer is the processed image, but the processed image is not particularly limited as long as it includes a surface image of some object.

  As described above, the image processing method and the image processing device according to the present invention have an effect that an image subjected to appropriate image processing can be obtained in the region of the edge portion of the object surface image, It is useful as an image processing method and an image processing apparatus for processing an object surface image composed of grayscale values in pixel units, and is suitable for a surface inspection apparatus.

DESCRIPTION OF SYMBOLS 10 Wafer 20 Support mechanism 21 Table 22a, 22b Support leg 30 Camera unit 31 Illumination unit 50 Base 55 Frame 60 Carriage 70 Shift movement mechanism 71 Stage 72 Drive part 100 Processing unit 111 Operation unit 112 Display unit 120 Drive control unit 200 Movement mechanism

Claims (30)

An image acquisition step configured to acquire an image to be processed including an object surface image, which is configured by a grayscale value in pixel units;
An edge image specifying step for specifying, as an edge image, an image of a region that includes an edge line of the object surface image set in the processed image and has a predetermined width in a direction orthogonal to the edge line;
A conversion step of converting the edge image into a rectangular image having a width corresponding to the predetermined width;
An image processing method comprising: an image processing step for performing one-dimensional image processing for sequentially processing the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image.   The image processing method according to claim 1, further comprising: an inverse conversion step of inversely converting the rectangular image subjected to the one-dimensional image processing into an image of an original region to generate a processed edge image.   In the edge image specifying step, an image of a region having a first predetermined width in an inner direction of the object surface image of the edge line and a second predetermined width in an outer direction of the object surface image of the edge line is obtained. The image processing method according to claim 1, wherein the image processing method is specified as the edge image. The converting step includes
A position conversion step of obtaining a position in the edge image corresponding to each pixel of the rectangular image;
4. A step of determining a gray value of a pixel of the rectangular image corresponding to each position in the acquired edge image based on a gray value of each pixel in the edge image. An image processing method described in 1. A conversion table representing a position in the edge image corresponding to each pixel of the rectangular image is created in advance,
The image processing method according to claim 4, wherein the position conversion step acquires a position in the edge image corresponding to each pixel of the rectangular image using the conversion table. The inverse transformation step includes
A position reverse conversion step of acquiring a position in a rectangular image corresponding to each pixel of the image of the original region;
Based on the gray value of each pixel in the rectangular image subjected to the one-dimensional image processing, the gray value of the pixel in the image of the original region corresponding to each position in the acquired rectangular image is determined, The image processing method according to claim 2, further comprising a step of generating a processed edge image. Create in advance an inverse conversion table representing the position in the rectangular image corresponding to each pixel of the image of the original area,
The image processing method according to claim 6, wherein the position reverse conversion step acquires a position in the rectangular image corresponding to each pixel of the image of the original region using the reverse conversion table. The image acquisition step acquires an image including a surface image of a disk-shaped semiconductor wafer as the processed image,
2. The edge image specifying step specifies all or a part of an image of a ring-shaped region having a predetermined width in a direction orthogonal to a circular line along the edge of the semiconductor wafer surface image as an edge image. Image processing method. The image acquisition step acquires an image including a surface image of a disk-shaped semiconductor wafer as the processed image,
The edge image specifying step specifies, as an edge image, an image of all or part of a ring-shaped region having a predetermined width in a direction orthogonal to a circular line along the edge of the semiconductor wafer surface image,
3. The inverse transformation step generates the processed edge image by inversely transforming the rectangular image subjected to the one-dimensional image processing into an image of all or a part of an original ring-shaped region. Image processing method.   The edge image specifying step specifies, as an edge image, an image of all or part of a ring-shaped region having a first predetermined width inside the circular line and a second predetermined width outside the circular line. The image processing method according to claim 8 or 9.   In the conversion step, each of the edge images is set at a predetermined interval along one of the two division lines parallel to the edge image of the object image with the edge line in between. The image according to claim 1, wherein the image is converted into a rectangular image composed of pixels arranged in a rectangular shape corresponding to each position set at a predetermined interval from the position and each normal to the other division line on a normal line orthogonal to the division line. Processing method.   In the conversion step, the edge image is set at a predetermined interval along a longer lane line between two lane lines parallel to the edge image with the edge line of the object image in between. 12. The image is converted into a rectangular image composed of pixels arranged in a rectangle corresponding to each position set at a predetermined interval from each position and each pixel to a shorter dividing line on a normal line orthogonal to the dividing line. Image processing method. A conversion table is prepared in advance for associating each position along the partition line in the edge image with one column of pixels of the rectangular image associated with all positions set on the normal passing through the position. Aside,
The image processing method according to claim 11, wherein the conversion step includes a step of determining a position in the edge image corresponding to each pixel of the rectangular image using the conversion table. An image acquisition step of acquiring an image to be processed partitioned by two parallel partition lines;
Each position of the image to be processed is set at a predetermined interval along one of the two parallel partition lines of the image to be processed, and a normal line orthogonal to the partition line from each position. An image processing method comprising a conversion step of converting to a rectangular image composed of pixels arranged in a rectangular pattern corresponding to positions set at predetermined intervals up to the other partition line.   In the converting step, the image to be processed is set at predetermined intervals along the longer lane line of the two parallel lane lines of the image to be processed and the lane line from each position. The image processing method according to claim 14, wherein the image is converted into a rectangular image made up of pixels arranged in a rectangular pattern corresponding to each position set at a predetermined interval up to a shorter dividing line on a normal line orthogonal to. Create in advance a conversion table for associating each position along the partition line in the image to be processed with one column of pixels of the rectangular image associated with all positions set on the normal passing through the position. Aside,
16. The image processing method according to claim 14, wherein the conversion step includes a step of determining a position in the image to be processed corresponding to each pixel of the rectangular image using the conversion table. Photographing means for photographing the surface of the object;
A processing unit for processing a photographed image obtained by the photographing means,
The processing unit is
An image acquisition unit configured to obtain a processed image including a surface image of the object, which is configured by a grayscale value in pixel units;
Edge image specifying means for specifying, as an edge image, an image of a region including a border line of the object surface image set in the processed image and having a predetermined width in a direction orthogonal to the border line;
Conversion means for converting the edge image into a rectangular image having a width corresponding to the predetermined width;
An image processing apparatus comprising: an image processing unit that performs one-dimensional image processing for sequentially processing the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image.   The image processing apparatus according to claim 17, further comprising: an inverse conversion unit configured to inversely convert the rectangular image subjected to the one-dimensional image processing into an image of an original region to generate a processed edge image. The converting means includes
Position conversion means for acquiring a position in the edge image corresponding to each pixel of the rectangular image;
The image according to claim 17 or 18, further comprising means for determining a gray value of a pixel of the rectangular image corresponding to each position in the acquired edge image based on a gray value of each pixel in the edge image. Processing equipment. Storage means for storing a conversion table representing a position in the edge image corresponding to each pixel of the rectangular image;
The image processing apparatus according to claim 19, wherein the position conversion unit acquires a position in the edge image corresponding to each pixel of the rectangular image using the conversion table. The photographing means photographs the surface of a disk-shaped semiconductor wafer,
The image acquisition means acquires an image including the semiconductor wafer surface image as the processed image,
The edge image specifying means specifies all or part of an image of a ring-shaped region having a predetermined width in a direction orthogonal to a circular line along the edge of the semiconductor wafer surface image as an edge image;
The image according to claim 17, wherein the inverse conversion unit generates a processed edge image by inversely converting the rectangular image subjected to the one-dimensional image processing into an image of all or a part of an original ring-shaped region. Processing equipment.   The edge image specifying means specifies, as an edge image, an image of all or part of a ring-shaped region having a first predetermined width inside the circular line and a second predetermined width outside the circular line. The image processing apparatus according to claim 21.   The converting means sets the edge image at predetermined intervals along one of the two dividing lines parallel to the edge image of the object image across the edge line. 18. The image according to claim 17, wherein the image is converted into a rectangular image composed of pixels arranged in a rectangular shape corresponding to each position set at a predetermined interval from the position and each normal to the other division line on a normal line orthogonal to the division line. Processing equipment.   In the conversion step, the edge image is set at a predetermined interval along a longer lane line between two lane lines parallel to the edge image with the edge line of the object image in between. 12. The image is converted into a rectangular image composed of pixels arranged in a rectangle corresponding to each position set at a predetermined interval from each position and each pixel to a shorter dividing line on a normal line orthogonal to the dividing line. Image processing apparatus. A conversion table that associates pixels for one column of the rectangular image associated with all positions set on the normal passing through the position at each position along the partition line in the edge image is stored. Having storage means;
25. The image processing apparatus according to claim 23, wherein the conversion means includes means for determining a position in the edge image corresponding to each pixel of the rectangular image using the conversion table. Photographing means for photographing the surface of the object;
A processing unit for processing a photographed image obtained by the photographing means,
The processing unit is
Image acquisition means for acquiring an image to be processed partitioned by two parallel partition lines from the captured image;
Each position of the image to be processed is set at a predetermined interval along one of the two parallel partition lines of the image to be processed, and a normal line orthogonal to the partition line from each position. And a conversion means for converting into a rectangular image composed of pixels arranged in a rectangular pattern corresponding to each position set at predetermined intervals up to the other partition line.   The converting means is configured to convert the image to be processed from each position set at a predetermined interval along a longer lane line of the two parallel lane lines of the image to be processed and the lane line from each position. 27. The image processing apparatus according to claim 26, wherein the image processing apparatus converts the image into a rectangular image composed of pixels arranged in a rectangular pattern corresponding to each position set at a predetermined interval up to a shorter dividing line on a normal line orthogonal to. A conversion table for associating pixels in one column of the rectangular image associated with all positions set on the normal passing through the position at each position along the partition line in the processed image is stored. Having storage means;
28. The image processing apparatus according to claim 26, wherein the conversion means includes means for determining a position in the processed image corresponding to each pixel of the rectangular image using the conversion table. Photographing means for photographing the surface of the object;
A processing unit that detects a defect on the object surface by processing a photographed image obtained by the photographing means;
The processing unit is
An image acquisition unit configured to obtain a processed image including a surface image of the object, which is configured by a grayscale value in pixel units;
Edge image specifying means for specifying, as an edge image, an image of a region including a border line of the object surface image set in the processed image and having a predetermined width in a direction orthogonal to the border line;
Conversion means for converting the edge image into a rectangular image having a width corresponding to the predetermined width;
First image processing means for performing one-dimensional image processing for sequentially processing the grayscale values of a plurality of pixels arranged in a direction orthogonal to the width direction of the rectangular image;
Inverse transform means for inversely transforming the rectangular image subjected to the one-dimensional image processing into an image of an original region to generate a processed edge image;
A main body image obtained by removing the edge image from the object surface image is subjected to two-dimensional image processing for sequentially processing gray values of a plurality of pixels arranged two-dimensionally to generate a processed object surface main body image. A second image processing means;
A surface inspection apparatus comprising defect specifying means for specifying a defective portion in the processed edge image and the processed main body image. 30. The surface inspection apparatus according to claim 29, wherein the imaging unit images a surface of a disk-shaped semiconductor wafer, and the processing unit detects a defect on the surface of the semiconductor wafer.
The image acquisition means acquires an image including a surface image of the semiconductor wafer as the processed image,
The edge image specifying means specifies all or part of an image of a ring-shaped region having a predetermined width in a direction orthogonal to a circular line along the edge of the semiconductor wafer surface image as an edge image;
The inverse conversion means generates a processed edge image by inversely converting the rectangular image subjected to the one-dimensional image processing into an image of all or part of the original ring-shaped region,
The second image processing unit is a surface inspection apparatus that generates a processed main body image by performing the two-dimensional image processing on the main body image obtained by removing the edge image from the semiconductor wafer surface image.

Images