JP2005127989A - Flaw detector and flaw detecting program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flaw detector or the like capable of detecting a flaw, foreign matter or the like having direction dependence with good precision for a short time. <P>SOLUTION: A film 9 being an object to be inspected is simultaneously irradiated with respective dark field illumination lights of an R dark field luminaire 1a, a G dark field luminaire 1b and a B dark field luminaire 1c from different directions and the dark field image of the object to be inspected is taken by a transmission image photographing camera 4 equipped with an RGB three-plate type imaging element and the color dark field image photographed by an image processing part 5 is separated into monochromatic image data of every color. Top hat conversion is adapted to the respective monochromatic image data to extract three flaw image data and one integrated flaw image data having flaw image data becoming the maximum value as the response value of each of pixels is extracted from three flaw image data with respect to each of pixels and the flaw or foreign matter such as dust or the like applied to the film 9 is detected on the basis of the integrated flaw image data. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a flaw detection device and a flaw detection program, and more particularly, to a flaw detection device and a flaw detection program for detecting a foreign matter such as a flaw and dust on a subject.

  Various devices (scratch detection devices) for detecting foreign matter such as scratches and dust on a subject such as a film have been conventionally proposed. For example, foreign substances such as dust often have a property of scattering illumination light and are detected as brighter spots than the surroundings. Therefore, a technique that employs a dark field illumination method for a flaw detection device is known.

  For example, Japanese Examined Patent Publication No. 6-43968 describes a technique for determining, as a foreign object, a black spot detected by transmitted bright field illumination and a white spot detected by transmitted dark field illumination.

  However, it is necessary to pay attention to the direction of scattered light in order to detect a foreign object, a scratch, or the like as a brighter point than the surroundings by such a dark field illumination method. This is because some scratches are detected from any direction, and some scratches are detected only from a specific direction.

  In order to detect the direction-dependent flaws such as the latter with high accuracy, there has been proposed an apparatus devised so as to eliminate blind spots depending on directions by irradiating illumination light from a plurality of directions. As an example using such a technique, in Japanese Patent No. 2705564, light having an angle of 10 ° to 40 ° with the normal line of a transparent glass substrate as an object is simultaneously transmitted from at least three directions. A transparent glass substrate defect detection device adapted to irradiate is described. This is an excellent defect detection device that has been devised to eliminate blind spots.

  Japanese Patent Application Laid-Open No. 2002-310935 describes an appearance inspection method for calculating a luminance average for each pixel and synthesizing a plurality of captured images taken at different illumination positions.

However, the techniques described in the above-mentioned Japanese Patent Nos. 2705564 and 2002-310935 have a problem that the response signal of the direction-dependent defect is weakened due to light superposition and signal averaging. In view of this, a technique has been proposed in which light from a plurality of directions is not simply superimposed but is irradiated independently at different times and sequentially photographed. As an example of such a technique, Japanese Patent Application Laid-Open No. 7-103905 discloses three illumination units arranged at positions shifted by about 90 ° in the horizontal direction, two imaging units arranged therebetween, And a translucent glass product flaw detection device that is adapted to capture dark field images in different directions in time series in operation.
Japanese Examined Patent Publication No. 6-43968 Japanese Patent No. 2705564 JP 2002-310935 A JP 7-103905 A

  As described above, in the device described in the above-mentioned Japanese Patent No. 2705564, since the light in a plurality of directions is simply superimposed, the response signal such as a direction-dependent scratch is weakened, and the detection sensitivity is lowered. There is a problem of end. This is the same even when a ring-shaped illumination light source is used.

  Moreover, in the thing of the said Unexamined-Japanese-Patent No. 2002-310935, since the average of a brightness | luminance is used, the response signal of a direction dependence defect is weakened similarly to the case of the said light superimposition. As a result, the detection sensitivity of scratches is lowered.

  On the other hand, in the device described in Japanese Patent Application Laid-Open No. 7-103905, since a plurality of images are taken in time series, the shooting time becomes long and the processing is slow. Therefore, it is difficult to apply this technique to a film digitizing apparatus or the like that is desirably equipped with a high-speed flaw detection function and a flaw correction function.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a flaw detection device and a flaw detection program capable of accurately detecting a direction-dependent flaw in a short time.

  In order to achieve the above object, a flaw detection apparatus according to a first invention is a flaw detection apparatus for detecting a foreign object or a flaw on a subject, and has a plurality of types of wavelength ranges with respect to the subject. Illuminating means for irradiating illumination light, and imaging means for capturing a subject illuminated by the illuminating means as a dark field image.

  Further, the scratch detection apparatus according to the second invention is the scratch detection apparatus according to the first invention, wherein the illuminating means is such that the overlapping region where the wavelength regions overlap is a sufficiently narrow region with respect to each wavelength region. Irradiates illumination light in a plurality of types of wavelength ranges.

  Further, the scratch detection apparatus according to the third invention is the scratch detection apparatus according to the second invention, wherein the plurality of types of wavelength ranges are a wavelength range expressing red, a wavelength range expressing green, and a wavelength expressing blue. Including at least two wavelength regions.

  The wound detection apparatus according to a fourth aspect of the present invention is the scratch detection apparatus according to the first aspect, wherein the illuminating means has a plurality of light sources respectively corresponding to the plurality of types of wavelength ranges in directions different from each other with respect to the subject. The imaging means is configured to be capable of simultaneously imaging a subject illuminated with illumination light of a plurality of types of wavelength ranges simultaneously irradiated from the plurality of light sources. It has been done.

  A scratch detection apparatus according to a fifth aspect of the present invention is the scratch detection apparatus according to the fourth aspect of the present invention, wherein the subject is in a direction along a plane normal to the imaging direction by the imaging means with respect to the imaging means. In at least one of the plurality of light sources, the direction in which the direction of illuminating the subject is projected onto the plane is a direction independent of the direction of transport.

  A scratch detection apparatus according to a sixth aspect of the present invention is the scratch detection apparatus according to the fifth aspect of the present invention, wherein the plurality of types of wavelength ranges include at least a wavelength range that expresses red, and illumination light in the wavelength range that expresses red is used. The light source to be irradiated is a light source in which the direction in which the direction of illuminating the subject is projected onto the plane is a direction independent of the transport direction.

  A scratch detection apparatus according to a seventh aspect of the present invention is the scratch detection apparatus according to the fourth aspect of the present invention, wherein the subject is in a direction along a plane normal to the imaging direction of the imaging means with respect to the imaging means. In at least one of the plurality of light sources, the direction in which the direction of illuminating the subject is projected onto the plane is a direction dependent on the direction of the conveyance.

  The scratch detection apparatus according to an eighth aspect of the present invention is the scratch detection apparatus according to the seventh aspect of the invention, wherein the plurality of types of wavelength ranges include at least a wavelength range that expresses red, and illumination light in the wavelength range that expresses red is used. The light source to be irradiated is a light source in which the direction in which the direction of illuminating the subject is projected on the plane is a direction dependent on the transport direction.

  A scratch detection apparatus according to a ninth aspect of the present invention is the scratch detection apparatus according to the fourth aspect of the present invention, further comprising second illumination means for irradiating a subject with white illumination light, wherein the imaging means is The object illuminated by the two illumination means can be captured as a bright field image.

  A flaw detection apparatus according to a tenth aspect of the invention is the flaw detection apparatus according to the fourth aspect of the invention, wherein the imaging means includes a color camera that can image a subject illuminated with illumination light of the plurality of types of wavelength ranges. Yes.

  The flaw detection apparatus according to an eleventh aspect of the invention is the flaw detection apparatus according to the ninth aspect of the invention, wherein the imaging means is illuminated with the subject illuminated with illumination light of the plurality of types of wavelength ranges and the white illumination light. And a color camera capable of imaging the subject.

  A flaw detection device according to a twelfth invention is the flaw detection device according to the fourth invention, wherein dark field images for each of the plurality of types of wavelength regions are extracted from the image data obtained by the imaging means. A dark field image recording means for recording, and a dark field image image obtained by calculating a luminance average value of the dark field image recorded in the dark field image recording means and normalizing the dark field image by the luminance average value A flaw image extraction processing means for extracting local variation data of pixel values based on data as flaw image data, and a pixel based on flaw image data for each of the plurality of types of wavelength regions extracted by the flaw image extraction processing means. A flaw image integration processing unit that extracts a maximum value for each to generate one integrated flaw image data, and a pixel value based on the flaw image data generated by the flaw image integration processing unit A wound determination unit determines that foreign matter or a scratch or the like attached to the object when equal to or greater than a predetermined threshold as compared, is obtained by further comprising a.

  A flaw detection device according to a thirteenth invention is the flaw detection device according to the ninth invention, wherein the imaging means captures the plurality of wavelengths from image data obtained by imaging the subject illuminated by the illumination means. A dark field image recording unit that extracts and records a dark field image for each area, and image data obtained by imaging the subject illuminated by the second illumination unit by the imaging unit as a bright field image Bright field image recording means for recording and the average brightness value of the dark field image recorded in the dark field image recording means, and the dark field image is normalized by the brightness average value to generate normalized dark field image data In addition, the brightness average value of the bright field image recorded in the bright field image recording means is calculated, and the bright field image is normalized by the brightness average value to generate normalized bright field image data. Flaw image extraction for further calculating difference image data obtained by subtracting the normalized bright field image data from the normalized dark field image data and extracting local change amount data of pixel values as wound image data based on the difference image data Processing means and flaw image integration for generating one integrated flaw image data by extracting the maximum value for each pixel based on the flaw image data for each of the plurality of types of wavelength ranges extracted by the flaw image extraction processing means A pixel value is compared with a predetermined threshold value based on the flaw image data generated by the processing unit and the flaw image integration processing unit, and when the pixel value is equal to or greater than the predetermined threshold value, it is determined that the object is a foreign object or a flaw on the subject And a scratch determination means.

  A flaw detection device according to a fourteenth invention is a flaw detection device for detecting a foreign matter or a flaw on a subject, and a light source that emits substantially white illumination light, which is visible light, is provided on the subject. Illumination means arranged to irradiate illumination light from a direction substantially orthogonal to the surface, and at least configured to capture a subject illuminated with illumination light from the light source as a dark field image And one dark field image camera.

  A flaw detection device according to a fifteenth aspect of the present invention is the flaw detection device according to the fourteenth aspect, wherein the flaw detection device is configured to capture a subject illuminated with illumination light from the light source as a bright field image. A camera is further provided.

  A flaw detection apparatus according to a sixteenth aspect of the present invention is the flaw detection apparatus according to the fourteenth aspect, wherein the subject is directed to the illumination means in one direction along a plane normal to the illumination direction of the illumination means. In at least one of the dark field image cameras, the direction in which the direction in which the subject is imaged is projected onto the plane is a direction independent of the direction of the conveyance. .

  A flaw detection apparatus according to a seventeenth aspect of the present invention is the flaw detection apparatus according to the fourteenth aspect, wherein the subject is directed to the illumination means in one direction along a plane whose normal is the illumination direction of the illumination means. In at least one of the dark field image cameras, a direction in which the direction in which the subject is imaged is projected onto the plane is a direction dependent on the direction of the conveyance. .

  A flaw detection device according to an eighteenth invention is the flaw detection device according to the fifteenth invention, wherein a plurality of the dark field image cameras are provided, and the plurality of dark field image cameras have a direction in which the subject is imaged. A dark-field image recording unit configured to record a plurality of dark-field image data in different imaging directions obtained by imaging by the plurality of dark-field image cameras, and for the bright-field image Bright field image recording means for recording bright field image data obtained by imaging by the camera, and a brightness average value of the dark field image recorded in the dark field image recording means, and calculating the brightness average value based on the brightness average value. Normalized dark field image data is generated by normalizing the field image, and the brightness average value of the bright field image recorded in the bright field image recording unit is calculated, and the bright field image is calculated based on the brightness average value. To generate normalized bright-field image data, further calculate difference image data obtained by subtracting the normalized dark-field image data from the normalized dark-field image data, and locally calculate pixel values based on the difference image data Flaw image extraction processing means for extracting a large amount of change data as flaw image data, and extracting a maximum value for each pixel on the basis of the plurality of flaw image data extracted by the flaw image extraction processing means. Scratch image integration processing means for generating flaw image data, and a pixel value is compared with a predetermined threshold based on the flaw image data generated by the flaw image integration processing means and the subject is attached when the pixel value is equal to or greater than a predetermined threshold Flaw determining means for determining that the object is a foreign object or a flaw.

  A flaw detection device according to a nineteenth invention is the flaw detection device according to the fourteenth invention, wherein the subject is a film-like object having a planar surface, and the dark field image camera is the subject. An optical means for forming an optical image of the above and an image pickup device for converting an optical image formed on the image pickup surface by the optical means into an electrical signal. The imaging device is arranged such that the imaging surface thereof is substantially parallel to the planar surface of the subject, and the optical means is configured such that the imaging surface and the planar surface of the subject are conjugated. The principal ray passing through the center of the imaging surface among the rays from the planar surface of the subject to the imaging surface intersects the optical axis of the optical means. It is comprised as follows.

  A flaw detection device according to a twentieth invention is the flaw detection device according to the fifteenth invention, wherein the subject is a film-like object having a planar surface, and the dark field image camera is the subject. An optical means for forming an optical image of the above and an image pickup device for converting an optical image formed on the image pickup surface by the optical means into an electrical signal. The imaging device is arranged such that the imaging surface thereof is substantially parallel to the planar surface of the subject, and the optical means is configured such that the imaging surface and the planar surface of the subject are conjugated. The principal ray passing through the center of the imaging surface among the rays from the planar surface of the subject to the imaging surface intersects the optical axis of the optical means. It is comprised as follows.

  The flaw detection apparatus according to a twenty-first aspect is the flaw detection apparatus according to the twentieth aspect, wherein the bright field image camera picks up the transmitted light of the subject illuminated by the illumination light from the light source as a bright field image. A second optical means for forming an optical image of the subject, and an optical image formed on the imaging surface by the second optical means. A second imaging device that converts the signal into a simple signal, and the second imaging device has an imaging surface substantially parallel to the planar surface of the subject. And the second optical means acts so that the imaging surface of the second imaging element and the planar surface of the subject have a conjugate relationship, The imaging of light rays from the planar surface to the imaging surface of the second imaging device Principal ray passing through the center of the surface is one that is configured so as to be substantially parallel to the optical axis of the second optical means.

  A flaw detection program according to a twenty-second invention is a flaw detection program for detecting a foreign object, a flaw, or the like attached to a subject. A first procedure for reading a visual field image, and a second procedure for recognizing a portion having a high luminance value among a plurality of dark field images read in the first procedure as a foreign matter or a flaw on a subject. This is a program to be executed.

  The flaw detection program according to a twenty-third invention is the flaw detection program according to the twenty-second invention, wherein the second procedure is a luminance average value of each dark-field image obtained by reading the plurality of dark-field images read in the first procedure. And a fourth procedure for extracting a top hat value, which is a high-frequency peak, from a local change amount in the vicinity of each pixel of a plurality of images standardized by the third procedure. And a fifth procedure for generating one image from a plurality of images using the maximum value selected from the top hat values of the plurality of images extracted in the fourth procedure, and A sixth procedure for selecting a pixel portion such as a foreign object or a flaw by comparing a top hat value in one image generated by the fifth procedure with a predetermined threshold value.

  The flaw detection program according to a twenty-fourth invention is the flaw detection program according to the twenty-second invention, wherein the second procedure is a luminance average value of each dark-field image obtained by reading the plurality of dark-field images read in the first procedure. , The eighth procedure for reading the bright field image of the subject illuminated by the illumination light from the light source, and the bright field image read in the eighth procedure. And a tenth procedure for subtracting the brightness value of the bright field image normalized in the ninth procedure from each dark field image normalized in the seventh procedure. And an eleventh procedure for extracting a top hat value, which is a high-frequency peak, from a local change amount in the vicinity of each pixel of a plurality of images of the subtraction result obtained in the tenth procedure, and the eleventh procedure. Multiple images extracted in A twelfth procedure for generating a single image from a plurality of images using a maximum value selected from among the top hat values selected, and a single image generated by the twelfth procedure. And a thirteenth procedure for selecting a pixel portion such as a foreign object or a flaw by comparing the top hat value with a predetermined threshold value.

  According to the flaw detection apparatus and flaw detection program of the present invention, it is possible to accurately detect flaws having direction dependency in a short time.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIGS. 1 to 3 show Embodiment 1 of the present invention, and FIG. 1 is a perspective view and FIG.

  In this embodiment, the object is a substantially transparent film-like object having optical transparency. That is, this is an embodiment relating to a flaw detection device for detecting foreign matters such as flaws and dust on the film 9.

  First, with reference to FIG. 1, the outline | summary of a structure of this flaw detection apparatus is demonstrated.

  A direction perpendicular to the main surface of the film 9 spread in a flat shape is the Z-axis direction, one direction in the main surface of the film is the X-axis direction, and a direction perpendicular to the X-axis direction in the main surface is the Y-axis direction. And

  At this time, an illuminating means for performing dark field illumination is disposed in a position facing the film 9, for example, a spatial area in which Z is positive, and the illuminating means is illuminated in a spatial area in which Z is negative. An image pickup means for picking up an image of the film 9 is arranged.

  The illumination means includes an R dark field illumination device 1a that is a light source that emits illumination light in a wavelength range that represents red (R), and a G dark field illumination that is a light source that emits illumination light in a wavelength region that represents green (G). The device 1b is configured to include a B dark field illumination device 1c that is a light source that emits illumination light in a wavelength range expressing blue (B), and each of these dark field illumination devices 1a, 1b, and 1c includes: As shown in FIGS. 1 (A) and 1 (B), the optical axis for irradiating illumination light is disposed at three equal positions around the Z axis so that the elevation angle is 45 degrees from the XY plane. That is, the R dark field illumination device 1a is arranged with an elevation angle of 45 degrees from the XY plane so that the line when the optical axis 2a of the illumination light is projected onto the XY plane coincides with the X axis (azimuth 0 degree). Yes. Also, the G dark field illumination device 1b is 45 from the XY plane so that the line when the optical axis 2b of the illumination light is projected onto the XY plane coincides with a line having an angle of 120 degrees from the X axis (azimuth 120 degrees). It is arranged with an elevation angle of degrees. Then, the B dark field illumination device 1c is 45 from the XY plane so that the line when the optical axis 2c of the illumination light is projected onto the XY plane coincides with a line having an angle of 240 degrees from the X axis (azimuth 240 degrees). It is arranged with an elevation angle of degrees. Each of the dark field illumination devices 1a, 1b, and 1c irradiates illumination light in the R, G, and B wavelength regions such that the overlapping region where the wavelength regions overlap is a sufficiently narrow region with respect to each wavelength region. Is.

  Further, as described above, the imaging means is configured to have a transmission image photographing camera 4 for digitizing an image of a foreign matter such as a scratch or dust attached to the film 9 illuminated from the back. The photographing camera 4 is disposed at a negative position on the Z axis so that the photographing optical axis coincides with the Z axis. Furthermore, the positional relationship between the transmission image photographing camera 4 and the film 9 as the subject is such that the Z axis that coincides with the photographing optical axis coincides with the normal line set at the center of the film 9. ing. In this embodiment, the transmission image capturing camera 4 is configured as a color camera including an RGB color three-plate image sensor.

  The transmission image photographing camera 4 is connected to the image processing unit 5. The image processing unit 5 is for performing processing for detecting foreign matters such as scratches and dust on the film 9 based on image data captured and output by the transmission image capturing camera 4.

  The operation of such a flaw detection device is as follows.

  First, from the RGB dark field illumination devices 1a, 1b, and 1c, the red illumination light indicated by the optical axis 2a, the green illumination light indicated by the optical axis 2b, and the blue illumination light indicated by the optical axis 2c are applied to the film. Simultaneously illuminate 9 toward obliquely.

  These illumination lights are transmitted as they are when there are no scratches or foreign matters on the film surface of the transparent film 9. On the other hand, when there are foreign matters such as scratches and dust on the film 9, the illumination light is scattered by these to become scattered light 3, and a part of the scattered light 3 enters the transmission image photographing camera 4. .

  The incident scattered light 3 is captured as an image related to each wavelength range of each illumination light by the RGB color three-plate image sensor of the transmission image capturing camera 4.

  Thus, by separating the dark field images captured by the three image sensors as image data for each wavelength, three monochrome images detected by the dark field illumination light in each direction can be extracted.

  Thus, the film 9 illuminated simultaneously from a plurality of spatially different positions is incident on one camera as an optical image, and image data is acquired simultaneously. Thereafter, the photographed image is input to the image processing unit 5 and is separated into three images according to the wavelength range, and a pixel having a flaw is determined.

  Next, FIG. 2 is a block diagram mainly showing a schematic configuration of the image processing unit.

  The image processing unit 5 includes a dark field image acquisition unit 11 that captures color image data input from the transmission image capturing camera 4, and the color image data acquired by the dark field image acquisition unit 11 as R image data, Dark field image recording units 12a, 12b, and 12c, which are dark field image recording units that record G image data and B image data separately as monochrome image information for each wavelength region, and these dark field image recording units 12a, 12a, And a flaw detection processing unit 13 that detects flaws, foreign matter, and the like based on images recorded in 12b and 12c.

  As described above, the images recorded in the dark field image recording units 12a, 12b, and 12c are R, G, and B dark field images illuminated in different directions from different directions, for each color. This is monochrome image information that is separated and digitized into luminance information in each wavelength region.

  Next, the flaw detection processing unit 13 calculates a response signal (a signal captured by the transmission image capturing camera 4) of each pixel based on each information recorded in the dark field image recording units 12a, 12b, and 12c. The flaw image extraction processing means 14a, 14b, 14c, which are flaw image extraction processing means for extracting the flaw images, and the flaw images extracted by the flaw image extraction processing sections 14a, 14b, 14c Since three monochrome images are picked up corresponding to the maximum response value (in this example, RGB), there are three response values at one pixel position, and the maximum response value is adopted. ) To obtain a flaw image integrated into one flaw image integration processing unit 15 as flaw image integration processing means, and a flaw position based on the flaw image generated by this flaw image integration processing unit 15 It determines that the wound determination unit serving wound determination unit 16 is configured to have a.

  Next, FIG. 3 is a flowchart showing a flaw detection process performed by the image processing unit 5.

  When the processing shown in FIG. 3A is started, first, the dark field image acquisition unit 11 reads dark field image data related to RGB dark field illumination from three different directions, and reads the read image. Recording is performed in the dark field image recording units 12a, 12b, and 12c, respectively (step S1).

  Next, the flaw image extraction processing units 14a, 14b, and 14c calculate the average luminance values of the three dark field images, respectively, and normalize the three dark field images by the calculated average luminance values, thereby normalizing the dark field. Image data is created (step S2).

  Then, the wound image integration processing unit 15 performs top hat (TH) conversion (top hat processing) on the three normalized dark field image data generated by the wound image extraction processing units 14a, 14b, and 14c. Then, a wound TH image is generated (step S3).

  Here, the top hat process includes two steps as shown in FIG.

  That is, first, an opening process which is one of basic operations in morphology is performed on the normalized dark field image data (step S11). Roughly speaking, this is a process of removing a shape portion smaller than the structuring element when performing the opening process.

  Next, the top hat transform value is calculated by subtracting the brightness value after the opening process from the brightness value of the original image before the opening process (step S12). Thus, generally speaking, only a shape portion smaller than the structuring element is extracted.

  By performing such top hat processing for each of the three standardized dark field image data, three flaw TH images are generated.

  Here, the description has been made assuming that the convex portion of the luminance is detected, but it may be configured to detect the concave portion of the luminance. For this purpose, top hat conversion may be performed in which the above-described opening process is replaced with a closing process.

  Returning to the process shown in FIG. 3A again, the process of selecting the largest top hat value among the top hat values at the same pixel position in the three scratch TH images as the top hat value of the pixel, By performing the processing for all the pixels, one flaw maximum response image data is generated and used as the integrated flaw image data (step S4).

  Here, if processing such as simply adding the three images generated by the flaw image extraction processing units 14a, 14b, and 14c or taking an average value is performed, a flaw having a direction dependency may be used. In some cases, the flaw response signal is weakened, and the contrast of the flawed portion is lowered, making detection difficult. Therefore, the images are synthesized by selecting the maximum response value as the pixel value instead of simply adding the images related to the illumination from the three directions.

  Next, the scratch determination unit 16 compares the maximum scratch response image data obtained in the process of step S4 with a threshold value that has been adjusted in advance (step S5). Is determined as a flaw pixel, and the position information of the pixel is recorded as flaw detection image data in a memory (not shown) or the like (step S6).

  If this step S6 is completed or if it is determined in step S5 that it is equal to or less than the threshold value, it is determined whether or not the processing for all pixels has been completed (step S7). In step S5, the above-described processing is repeated for the next pixel.

  Thus, when it is determined in step S7 that the processing for all the pixels has been completed, since the scratch detection image data in which the pixel positions of the scratches on the film are recorded has been generated, The detection process ends.

  Note that the threshold value used in step S5 is a value in a standardized range between 0 and 1 with respect to the signal level corresponding to the intensity of scattered light due to scratches or dust on the subject to be measured. Is set. Generally, a value close to 0 is set as the threshold value when it is desired to detect a scratch with small scattered light (such as a shallow scratch or a fine scratch) with high accuracy. However, since the scratch maximum response image data includes noise other than the response signal due to scattered light, it is necessary to measure the noise level in advance and set the threshold value to a value that does not pick up the noise. is there. If it is sufficient to detect only scratches with large scattered light (deep scratches, large scratches, etc.), it is possible to suppress picking up fine scratches by setting the threshold value to a larger value. And the processing time can be shortened.

  In addition, as the structuring element for performing the opening process in step S11, an element having an appropriate shape or size may be selected in accordance with the shape or size of a scratch or foreign object to be detected. In general, a small structural element is used when it is desired to detect even a small scratch. However, when it is not necessary to detect a very small scratch, a structural element having an appropriate size is used.

  In the above description, scratches and foreign objects are detected by top-hat conversion. However, the present invention is not limited to this, and other image processing or the like may be used to detect scratches or foreign objects.

  Further, in each of the above-described embodiments, the dark field illumination light is irradiated from three directions. However, the present invention is not limited to this, and the irradiation may be performed from two, four or more directions. I do not care. If the direction in which the dark field illumination light is irradiated is increased, a scratch image with fewer undetected scratches can be obtained.

  In the above description, scratches, foreign matter, and the like are detected using light in the three wavelength ranges of R, G, and B. As described above, only two of these colors are used. You may make it detect, and you may make it detect from more directions using the light of a multicolor wavelength range so that wavelength ranges may not overlap. At this time, the light in the visible light region is not necessarily used, and it is needless to say that light in the infrared region or light in the ultraviolet region may be used.

  Furthermore, in this embodiment, an example using a camera equipped with an RGB color three-plate image pickup device has been described. However, the present invention is not limited to this. For example, a single-plate camera using an RGB three-primary color mosaic filter or the like is used. It doesn't matter. In this case, the resolution of the dark-field monochrome image is reduced, but the pixels of each color are regularly arranged, so that image processing can be performed with simple processing, and the single-plate type is more inexpensive. It becomes possible to configure the apparatus.

  Alternatively, three monochrome cameras may be used as cameras for capturing a transmission image, and R, G, and B color filters may be arranged on the front side of the imaging element of each monochrome camera. In dark field imaging, these three monochrome cameras can capture a dark field image of a subject by R, G, and B dark field illumination, thereby obtaining a dark field image of each color from separate cameras. it can. Further, in bright field photography, one color original image can be obtained by synthesizing the images of the respective colors taken by the three cameras. However, in this case, the pixel position of a certain camera is used in order to correct a pixel position shift caused by a difference in imaging optical system of each camera, a difference and error in mounting position, a difference in mounting angle and error, etc. In other cameras, it is necessary to create a pixel correspondence map, a pixel correspondence table, and the like indicating which pixel position corresponds to, and perform calibration processing and inter-pixel correction processing.

  According to the first embodiment, since the dark field images having different orientations are obtained separately, the maximum response value of the wound is calculated and used for detection. It becomes possible to detect well without lowering the sensitivity. And by using the light of a different wavelength range, since the subject image illuminated from multiple directions can be acquired by one imaging, compared with the case where an image is acquired by three imaging. Images can be captured at high speed. Therefore, it can be satisfactorily applied to a film digitizing apparatus or the like that is desirably equipped with a high-speed scratch detection function or the like.

  4 to 6 show a second embodiment of the present invention, and FIG. 4 is a perspective view schematically showing the configuration of the flaw detection device. In the second embodiment, the same parts as those of the first embodiment are denoted by the same reference numerals and the description thereof is omitted, and only different points will be mainly described.

  In the present embodiment, a developed film 29 in which a plurality of images are continuously recorded in frame units (imaging area 29a in FIG. 4) is taken as an example, and foreign matter such as scratches and dust on the film 29 is taken as an example. This is an embodiment related to a flaw detection device for detecting the like. The flaw detection device of this embodiment also serves as a film digitizing device for converting an optical film image into a digital image.

  First, with reference to FIG. 4, the outline | summary of a structure of this flaw detection apparatus is demonstrated.

  Here, the direction perpendicular to the main surface of the film 29 developed in a planar shape is the Z-axis direction, the traveling direction that is one direction in the main surface of the film is the X-axis direction, and the X-axis direction in the main surface is The vertical direction is defined as the Y-axis direction.

  In this flaw detection device, the RGB dark field illumination devices 1a, 1b, 1c and the transmission image photographing camera 4 are arranged with the film 29 interposed therebetween as in the first embodiment, and further visible. A bright field illumination device 26 as a second illumination means for irradiating white light as illumination light is disposed.

  The bright field illumination device 26 is arranged such that the optical axis 27 of the illumination light coincides with the normal line (Z axis) set at the center of the imaging region 29a of the film 29 as the subject. 27 also coincides with the photographing optical axis of the transmission image photographing camera 4.

  Further, the image processing unit 5A that processes an image output from the transmission image capturing camera 4 has a slightly different configuration from the image processing unit 5 of the first embodiment, as will be described later with reference to FIG. Yes.

  Further, the film 29 is, for example, a general color print film, and is formed by laminating a base layer as a base of the film, an image recording layer for forming an image, and a protective layer for protecting the image recording layer. It is comprised by doing. The image recording layer further comprises a layer that is sensitive to red light, a layer that is sensitive to green light, and a layer that is sensitive to blue light, and the silver salt particles contained in each layer. Depending on the density, an image of each color is formed.

  Next, the operation of such a flaw detection device will be described.

  The film 29 is frame-fed by the transport device (not shown) in the X-axis direction, which is the longitudinal direction of the film 29, and the center of the shooting area 29a, which is a shooting frame, coincides with the shooting optical axis of the transmission image shooting camera 4. That is, it is positioned so as to be positioned on the optical axis 27 of the bright field illumination device 26.

  Next, the bright field illumination device 26 irradiates bright field illumination light (indicated by the optical axis 27) made of visible white light toward the photographing region 29 a of the film 29.

  The irradiation light is modulated in luminance by the image recorded in the photographing region 29 a and transmitted, and light 23 traveling straight as it enters the transmission image photographing camera 4. Thereby, the color image information recorded on the film 29 is acquired as a bright field image.

  Subsequently, from the RGB dark field illumination devices 1a, 1b, and 1c, red illumination light indicated by the optical axis 2a, green illumination light indicated by the optical axis 2b, and blue illumination light indicated by the optical axis 2c, The film 29 is irradiated simultaneously from an oblique direction.

  The dark field illumination light indicated by the optical axes 2a, 2b, and 2c is irradiated at a timing different from that of the bright field illumination light indicated by the optical axis 27, but the operation is the same as in the first embodiment. Yes, it is scattered by foreign matter such as scratches and dust on the film 29, and part of the scattered light is incident on the transmission image photographing camera 4 as light 23.

  The incident light 23 as the scattered light is captured as an image related to each wavelength range of each illumination light by the RGB color three-plate image sensor of the transmission image capturing camera 4.

  Thus, by separating the dark field images captured by the three image sensors as image data for each wavelength, three monochrome images detected by the dark field illumination light in each direction can be extracted.

  The four images (one color image and three monochrome images) acquired by the two imaging operations as described above are input to the image processing unit 5A and, as will be described later, the presence of scratches. The pixel to be determined is determined.

  Next, FIG. 5 is a block diagram mainly showing a schematic configuration of the image processing unit.

  The image processing unit 5A acquires a bright / dark field image acquisition unit 11A that captures bright field image data and dark field image data input from the transmission image capturing camera 4, and the bright / dark field image acquisition unit 11A. The dark field image recording units 12a, 12b, and 12c that separate the recorded dark field image data into R image data, G image data, and B image data, and record them as monochrome image information for each wavelength region; The bright field image recording unit 21 serving as a bright field image recording unit that records the bright field image data acquired by the field image acquisition unit 11A, the images recorded in the dark field image recording units 12a, 12b, and 12c, and the bright field A flaw detection processing unit 13 for detecting a flaw, a foreign object or the like based on the image recorded in the image recording unit 21, and a flaw detection image generated by the flaw detection processing unit 13 Based on the data it is configured to have a defect correction processing unit 22 corrects the original image data of the recorded film to the bright field image recording unit 21, a.

  The configuration of the flaw detection processing unit 13 is the same as that of the first embodiment.

  The capture timing generator 24 is for generating a signal that synchronizes the timing at which the light source emits light and the timing at which an image is captured. Status information and signals such as commands are transferred to the dark field image acquisition unit 11A.

  The light source switching unit 25 performs a process of alternately switching between the bright field illumination device 26 and the RGB dark field illumination devices 1a, 1b, and 1c in accordance with a signal from the capture timing generator 24.

  The signal from the capture timing generator 24 further serves as a trigger signal when image capture is performed by the bright / dark field image acquisition unit 11A, and a color image is generated according to the status signal related to the trigger signal. Is transferred to the bright field image recording unit 21 as it is, or is separated for each color as monochrome image information and transferred to the dark field image recording units 12a, 12b, and 12c.

  Next, FIG. 6 is a flowchart showing the contents of processing by the scratch detection device, and FIG. 6A is a flowchart showing the system operation of the scratch detection device.

  As shown in FIG. 6A, the operation of the system of the flaw detection apparatus is roughly classified into the following three processes.

  First, the frame 29 of the film 29 is moved to the photographing area by a transport device (not shown). Then, an image capturing process is performed to convert the optical image recorded on the film 29 into digitized electronic image data (step S21).

  Next, using the captured original film image and the three dark field images, a flaw detection process for detecting a flaw pixel position in the original film image is performed to create flaw detection image data (step S22). .

  Thereafter, when the flaw detection image data includes flaw information, a flaw correction process for correcting flaws in the original image data is performed based on the information (step S23).

  A series of processes from film frame advancement to image capture and image processing are repeated until the frames recorded on the film 29 are completed. However, it is also possible to perform film processing and image processing for each recorded image after performing film frame advance and image capture for all frames in advance. is there.

  Next, FIG. 6B is a flowchart showing details of the image capturing process in step S21.

  In the image capturing process, as described below, the image processing unit 5A sequentially performs two processes of image acquisition by bright field illumination and image acquisition by dark field illumination.

  First, the bright field illumination device 26 illuminates the shooting area 29a of the film 29 conveyed by the conveyance device. At this time, the dark field illumination devices 1a, 1b, and 1c are turned off, or all the dark field illumination light indicated by the optical axes 2a, 2b, and 2c is blocked. Only the bright field illumination light shown is performed (step S31).

  Next, the optical image of the photographing region 29a illuminated by the bright field illumination light is converted into an electronic image by the transmission image photographing camera 4 and output, and is captured by the bright / dark field image obtaining unit 11A of the image processing unit 5A. Then, it is recorded in the bright field image recording unit 21 (step S32). At this time, the switching timing of the illumination device and the capture timing are performed in synchronization with the status signal generated by the capture timing generator 24 as described above.

  Subsequently, the bright field illumination device 26 is turned off, or the bright field illumination light indicated by the optical axis 27 is blocked, and the dark field illumination devices 1a, 1b, and 1c are turned on simultaneously (step S33).

  After that, the optical image of the light 23 that is the scattered light of the dark field illumination light is converted into an electronic image by the transmission image photographing camera 4 and output, as described above, and the bright / dark field image of the image processing unit 5A. Captured by the acquisition unit 11A and recorded separately in the dark field image recording units 12a, 12b, and 12c (step S34).

  Next, FIG. 6C is a flowchart showing details of the flaw detection processing in step S22.

  When this scratch detection process is started, first, the bright field color image data recorded in the bright field image recording unit 21 and the three directions recorded in the dark field image recording units 12a, 12b, and 12c are recorded. The dark field monochrome image data is read (step S41).

  Here, when the film 29 is taken in the dark field, light scattering occurs in a portion such as a foreign object such as a scratch or dust, and scattered light that is stronger than other portions is acquired in the first embodiment. This is the same as the case of the transparent film as described above. However, in the case of the film 29 of this embodiment, since an image is recorded unlike the transparent film, the silver salt in the image recording layer inside the film 29 even if there is no scratch or foreign matter. Weak scattered light is generated by particles or the like, and an image in which an original image with low luminance is superimposed on a dark field image is acquired.

  Therefore, the flaw image extraction processing units 14a, 14b, and 14c calculate the average luminance values of the three dark field images, respectively, normalize the three dark field images based on the calculated luminance average values, and normalize the dark field. Image data is created (step S42).

  Since this original dark field image data includes the original image having a low luminance as described above, the RGB luminance average value of the color image data by the bright field illumination light is calculated, and the RGB Color image data is normalized by the average luminance value, and standardized bright field image data is created. Then, by subtracting the standardized bright field image data from the standardized dark field image data, a process for removing the original image having low luminance from the dark field image data is performed (step S43). Thus, the normalized dark field image data from which the original image with low luminance is removed becomes the difference image data. It should be noted that when this process is performed, 0 (zero) is used for a negative subtraction value.

  For the first time after performing the processing so far, three pieces of scratch-extracted image data from which scratch images from three directions are extracted are obtained.

  Subsequent processing is the same as the processing in steps S3 to S7 shown in FIG.

  FIG. 7 is a diagram illustrating a dark field image obtained by photographing each film with illumination from three directions and a processed image obtained by processing these dark field images in two ways.

  7A shows a dark field image obtained by illuminating the film from the 0 degree direction, FIG. 7B, the 120 degree direction, and FIG. 7C, the 240 degree direction.

  The white streak portion in these dark field images is the portion of the scratch image due to scattered reflection. As shown in the figure, the contrast of the flaw images changes depending on the direction of illumination. In particular, it can be seen that the flaw image portion on the lower left greatly changes depending on the direction of illumination.

  In addition, since the film is an image recorded, when the film is a complete transparent body, the recorded image has a low contrast (even thinly) in a portion that becomes completely dark (the entire dark field image surface). As observed).

  FIG. 7D is a diagram illustrating a case where a dark field image as shown in FIGS. 7A, 7B, and 7C is simply added (superposed), and a top hat is added to the added image. A processed image is shown as a result of binarization after the conversion (TH conversion) processing.

  FIG. 7E shows a dark field image as shown in FIG. 7A, FIG. 7B, and FIG. 7C, which is subjected to top hat conversion processing, and then synthesized as a maximum value image. And the processed image of the result of binarizing the maximum value image is shown.

  Comparing the processed image of FIG. 7D with the processed image of FIG. 7E, the processed image of FIG. 7D shows that the lower left flaw is not detected, but FIG. In the processed image of (E), it can be seen that all the flaws including the flawed portion at the lower left are well detected.

  As described above, since some of the response components of scattered light have an orientation dependency, if the image including such scratches is simply superimposed and added, the response component is weakened. . On the other hand, if the maximum value processing is performed as described above, the response signal is not weakened, so that all flaw images can be detected with higher accuracy.

  In addition, it is possible to detect a scratched part by performing a simple differentiation process instead of the above-described top hat conversion process. However, when performing the differentiation process, it may be necessary to deal with pixel shift. It may be necessary to adjust parameters such as setting differential values. On the other hand, the top hat conversion as described above has an advantage that a robust processing result can be obtained even by rough adjustment.

  The scratch correction process in step S23 is a process of repairing the scratch area based on the scratch position information (scratch detection image data) by the scratch correction processing unit 22, and the target frame recorded on the film 29 is repaired. Examples include restoration using the image information of the previous and subsequent frames, or interpolation by using pixel data in the vicinity of the scratch, but various known techniques can be used. Detailed description is omitted here.

  The film is likely to have many scratches in the running direction, but such scratches are often fine scratches with low contrast that do not affect the original image. Therefore, there is a tendency that the flaws in this direction are detected excessively, and a lot of time may be required for the flaw determination process. Therefore, by taking the arrangement of the dark field illumination device such that the line obtained by projecting the optical axis of the illumination light on the film surface does not become a direction orthogonal to the film traveling direction (more broadly, a direction independent of the film traveling direction) ( That is, for example, by taking the arrangement of a dark field illumination device such that a line obtained by projecting the optical axis of the illumination light on the film surface is a direction dependent on the film traveling direction (the same direction as the film traveling direction or the reverse direction)) The processing speed of flaw detection can be improved.

  On the other hand, the processing may be slow, but if you want to prevent the scratches from being undetected, the line that projects the optical axis of the illumination light on the film surface is orthogonal to the running direction of the film. If the dark field illumination device is arranged so as to be in such a direction (more broadly, a direction independent of the film traveling direction), it is possible to detect the scratch in the traveling direction with high accuracy, and to improve the performance of the scratch detection. Can be improved.

  Furthermore, the sensitivity of film flaw detection is not the same in any wavelength range, but actually the order of sensitivity is R> G> B. Therefore, the detection sensitivity of the flaw is improved by using only the light in the red (R) wavelength region as the dark field illumination light, rather than using the white light that is the average sensitivity of RGB. Therefore, by disposing a dark field illumination device that emits light in the red (R) wavelength region in a direction in which it is desired to increase detection sensitivity among a plurality of dark field images, it is possible to effectively improve the flaw detection performance. be able to. As a specific example of the effective arrangement, as described above, light in the red (R) wavelength region so that the line obtained by projecting the optical axis of the illumination light onto the film surface is in a direction orthogonal to the traveling direction of the film. It is possible to arrange a dark field illumination device that irradiates the light. By taking such an arrangement, it is possible to detect a scratch in the film running direction with higher accuracy.

  In the second embodiment, since the dark field illumination light and the bright field illumination light are switched, it takes a processing time to switch the illumination and take an image. Since the film image can be captured by a single camera, the apparatus can be realized with a very simple configuration even if a mechanism for switching the light source is necessary. it can.

  On the other hand, for example, in the case where the bright-field image camera and the dark-field image camera are separately installed, as described in the first embodiment, the imaging optical system used for detection is different. Not only that, but also magnification and image distortion (distortion) of the imaging optical system, differences and errors in the mounting position, differences and errors in the rotational position, and the like need to be measured in advance. Furthermore, using this measurement result, a camera for accurately associating which pixel position of the bright field image camera corresponds to the pixel position determined to be a scratch from the image of the dark field image camera Since pre-processing for creating a pixel correspondence map, a pixel correspondence table, and the like is required, a process using the correction information is also required during the flaw detection processing.

  However, in the second embodiment, since both the bright-field image and the dark-field image are captured by one transmission image capturing camera, such a problem does not occur and the scratches are corrected. A film scan image can be acquired in a relatively short time with an apparatus having a simple configuration.

  According to the second embodiment, the effects similar to those of the first embodiment described above can be obtained, the scratches on the film can be detected, and the image of the film can be acquired to correct the influence of the scratches. it can. Therefore, it can be satisfactorily applied to a film digitizing apparatus or the like that is desirably equipped with a high-speed flaw detection function and a flaw correction function.

  Even when an image is recorded on the film, it is possible to remove the influence of the image by subtracting the standardized bright field image from the standardized dark field image and perform accurate scratch measurement. is there.

  Furthermore, since the camera that captures the bright field image and the camera that captures the dark field image are shared, the film image can be converted into a digital image with a simple configuration. Since the dark field image is acquired by the light of each of the RGB wavelength ranges constituting the color original image, such a camera can be shared.

  FIGS. 8 to 12 show a third embodiment of the present invention, FIG. 8 is a perspective view showing an outline of the configuration of the scratch detection device from the bottom side, and FIG. 9 is a film surface, a lens, and an image pickup in the scratch detection device. FIG. 10 is a diagram showing an example of the positional relationship between the film surface, the lens, and the imaging surface, FIG. 11 is a block diagram mainly showing the outline of the configuration of the image processing unit, and FIG. These are the flowcharts which show each processing content by a wound detection apparatus.

  In the third embodiment, the same parts as those in the first and second embodiments are denoted by the same reference numerals, description thereof is omitted, and only different points will be mainly described.

  In the present embodiment, similar to the second embodiment described above, a developed film 29 in which a plurality of images are continuously recorded in frame units (imaging area 29a in FIG. 8) as an object is taken as an example. 29 is an embodiment relating to a flaw detection device for detecting foreign matter such as flaws and dust attached to 29. The flaw detection device of this embodiment also serves as a film digitizing device for converting an optical film image into a digital image.

  In the scratch detection apparatus of the first embodiment and the second embodiment described above, there is one camera as an imaging means, and a dark field illumination device as an illumination means that emits illumination light of a different wavelength is placed on the opposite side across the film. Although it was arranged in a plurality of directions, the flaw detection apparatus of Example 3 has one illuminating device as an illuminating means, and for a plurality of dark field images as imaging means on the opposite side across the film. The camera is arranged in a plurality of directions.

  With reference to FIG. 8, the outline | summary of a structure of this flaw detection apparatus is demonstrated.

  In Example 3, as in Examples 1 and 2 described above, the direction perpendicular to the main surface of the film 29 spread in a flat shape is the Z-axis direction, which is one direction within the main surface of the film 29. The longitudinal direction of the film 29 is the X-axis direction, the direction perpendicular to the X-axis direction in the main surface is the Y-axis direction, and the side where the illumination device is disposed is the positive direction of the Z-axis.

  That is, an illuminating device 31 that is an illuminating means is disposed in the positive direction of the Z axis that passes through the center of the photographing region 29a of the film 29 and is perpendicular to the film surface. The illuminating device 31 emits substantially white illumination light. It is a light source.

  Further, on the opposite side across the illumination device 31 and the film 29, for example, a bright-field image camera 35, which is a color camera including an RGB color filter, and three dark-field image cameras 36a, which are monochrome cameras, for example. , 36b, and 36c. Here, the bright-field image camera 35 is configured as a color camera because it is necessary to digitize the color information of the film 29. However, the dark-field image cameras 36a, 36b, and 36c Since it is sufficient that the intensity and its position can be acquired, it is configured as a monochrome camera.

  The bright field image camera 35 includes an imaging optical system 35a as a second optical unit for forming an optical image of the imaging region 29a, and an imaging surface 38a (see FIG. 9B) by the imaging optical system 35a. ) And an image pickup device 38 (see FIG. 9B) (second image pickup device) for photoelectrically converting the optical image formed on the optical image and outputting a color image signal. This is for digitizing the image recorded in the photographing area 29a. The bright field image camera 35 is disposed in the negative direction of the Z-axis direction so that the optical axis of the imaging optical system 35a coincides with the Z-axis described above.

  The dark field image cameras 36a, 36b, 36c include detection optical systems 37a, 37b, 37c, which are optical means for forming an optical image of the photographing region 29a as described later, and these detection optical systems 37a, An imaging element 39 (see FIG. 9B) for photoelectrically converting an optical image formed on each imaging surface 39a (see FIG. 9B) by 37b and 37c and outputting a monochrome image signal; , And is for acquiring the scattered light from the imaging region 29a of the film 29 as a dark field image.

  As shown in FIGS. 8 and 9, the dark field image cameras 36a, 36b, and 36c are configured so that the angle at which the scattered light from the film 29 is received is approximately 45 degrees with respect to the negative direction of the Z axis. Are arranged at three equal positions around the Z axis.

  That is, in the dark-field image camera 36a, the line when the principal ray of the scattered light 34a incident on the center of the imaging surface 39a of the image sensor 39 is projected on the XY plane coincides with the X axis (azimuth 0 degree). Thus, it is arranged with an angle of about 45 degrees with respect to the negative direction of the Z-axis. Further, in the dark field image camera 36b, the line when the principal ray of the scattered light 34b incident on the center of the imaging surface 39a is projected onto the XY plane coincides with a line having an angle of 120 degrees from the X axis (azimuth). 120 degrees), and is arranged at an angle of approximately 45 degrees with respect to the negative direction of the Z-axis. In the dark field image camera 36c, the line when the principal ray of the scattered light 34c incident on the center of the imaging surface 39a is projected onto the XY plane coincides with a line having an angle of 240 degrees from the X axis (azimuth). 240 degrees) with respect to the negative direction of the Z axis.

  The detection optical systems 37a, 37b, and 37c are configured as shift optical systems as will be described later, and the entire image area 29a developed in a planar shape on the film 29 is not defocused and distorted. The image is formed on the image pickup device 39 of each dark field image camera 36a, 36b, 36c without causing the above.

  The bright field image camera 35 and the dark field image cameras 36a, 36b, and 36c are connected to an image processing unit 5B. The image processing unit 5B uses the bright-field image output from the bright-field image camera 35 and the dark-field images output from the dark-field image cameras 36a, 36b, and 36c to cause scratches and dust on the film 29. And the like, and image processing such as correcting a flaw in a bright-field image based on the detected information is performed.

  Next, the operation of such a flaw detection device will be described.

  While the film 29 is frame-fed by the conveying device (not shown) in the X-axis direction, which is the longitudinal direction of the film 29, the center of the shooting area 29a, which is a shooting frame, becomes the shooting optical axis of the bright-field image camera 35. Positioned to match.

  Next, substantially white illumination light 32, which is visible light, is emitted from the illumination device 31 toward the photographing region 29 a of the film 29.

  The irradiation light is transmitted with its luminance modulated by an image recorded in the photographing region 29 a and transmitted straight as it is, and enters the bright field image camera 35. Thereby, the color image information recorded on the film 29 is acquired as a bright field image.

  At the same time, light scattering occurs due to foreign matters such as scratches and dust on the film 29. The light is scattered in various directions, and scattered light 34a, 34b, and 34c that form an angle of about 45 degrees with respect to the Z-axis among them is detected with respect to the detection optical systems 37a, 37b, and 37c. Each incident light is focused on the image sensor 39 of each dark field image camera 36a, 36b, 36c.

  As described above, since the four cameras of the bright field image camera 35 and the dark field image cameras 36a, 36b, and 36c are spatially separated from each other, the acquisition of the bright field image and the plurality of cameras are performed. It is possible to simultaneously acquire a dark field image of a direction.

  One bright field image and three dark field images captured in this manner are input to the image processing unit 5B, and determination of a pixel having a flaw, a flaw correction process, and the like are performed.

  Next, the configuration of the detection optical systems 37a, 37b, and 37c will be described in more detail with reference to FIG.

  First, as a configuration common to any of FIGS. 10A, 10 </ b> C, and 10 </ b> E, the optical axis of the illumination device 31 is, as described above, the center of the film surface F (the film 29. The center of the imaging region 29a) and parallel to the normal of the film surface F. Furthermore, since the arrangement of the dark field image cameras 36a, 36b, and 36c is described here, the lenses L0 to L3 and the imaging surface G as shown in the figure are dark field images due to scattered light from the film 29. Can be received, and the position where the light related to the bright field image is not incident, that is, the film surface F is inclined to a certain degree or more (that is, a predetermined level or more with respect to the normal line standing on the film surface F). (The direction of the angle) is as desired. Further, the imaging surface G and the focus surface P of the imaging element are in a conjugate relationship with each other.

  First, in FIG. 10A, the main surface of the lens L0 and the imaging surface G are parallel, and the optical axis of the lens L0 and the principal ray O passing through the center of the imaging surface G coincide with each other. It is an example which shows the structure of a general imaging optical system. In such an optical system, the film surface F and the focus surface P do not coincide with each other, but intersect with each other at an intersection (a three-dimensional intersection line).

  In the optical system having the arrangement as shown in FIG. 10A, even if the center of the film surface F coincides with the focus surface P, the image is in a range that falls within the depth of field, that is, in focus and has high contrast. As shown in the area sandwiched by the dotted lines in FIG. 10B, the range where the image can be obtained is a part of the image formation portion on the image pickup surface G (part along the horizontal direction near the center). . The upper and lower areas outside the area sandwiched by the dotted lines are not in focus and only an image with low contrast can be obtained. Furthermore, a rectangular observation region 29 on the film 29 also has a substantially isosceles trapezoidal shape in which the upper side is relatively reduced from the lower side as shown in FIG. 10B when this optical system is used. As a result, an image is formed on the imaging surface G, and geometric distortion of the image occurs.

  Therefore, in the optical system as shown in FIG. 10A, the region to be observed is small, the region falls within the depth of field of the optical system, and the distortion of the image can be almost ignored. Unless it can be done, it is not easy to use.

  Next, FIG. 10C shows that the main surface of the lens L1, the imaging surface G, and the focus surface P intersect with each other at an intersection (intersection line in three dimensions) and the film surface F that is the subject. It is an example which shows the structure of what is called a tilt optical system arrange | positioned so that the focus surface P may correspond. At this time, the optical axis of the lens L1 and the principal ray O passing through the center of the imaging surface G do not match.

  If an optical system having an arrangement as shown in FIG. 10C is used, an image with high contrast focused on the entire surface of the film surface F can be formed on the imaging surface G. However, even at this time, as shown in FIG. 10D, the geometric distortion of the image in which the rectangular object is imaged in a substantially isosceles trapezoidal shape still occurs.

  Therefore, the optical system as shown in FIG. 10C cannot be used unless it is assumed that geometric correction of an image obtained by imaging is performed, and a configuration such as a geometric correction circuit is necessary. As a result, the cost increases and the processing time becomes longer.

  Further, in FIG. 10E, the main surface of the lens L2, the imaging surface G, and the focus surface P are parallel to each other, and the film surface F and the focus surface P that are the subject coincide with each other. It is an example which shows the structure of what is called a shift optical system arrange | positioned. Therefore, the lens L2 acts so that the imaging surface G of the imaging device and the film surface F, which is the planar surface of the subject, are in a conjugate relationship. Also at this time, the optical axis of the lens L2 and the principal ray O passing through the center of the imaging surface G do not coincide.

  If an optical system having an arrangement as shown in FIG. 10E is used, an image with high contrast focused on the entire surface of the film surface F can be formed on the imaging surface G. Further, at this time, as shown in FIG. 10 (F), a rectangular subject is imaged in a rectangular shape, and no geometric distortion of the image resulting from the tilt of the optical system has occurred.

  Therefore, it can be said that the optical system as shown in FIG. 10E is suitable for use in the detection optical systems 37a to 37c of the dark field image cameras 36a to 36c.

  Thus, as shown in FIGS. 8 and 9, it is this shift optical system that is applied to the detection optical systems 37a to 37c.

  That is, the dark-field image cameras 36a, 36b, and 36c arranged around the photographing optical axis of the bright-field image camera 35 in a substantially circumferential shape are 120 degrees azimuth as shown in FIG. The detection optical systems 37a, 37b, and 37c are arranged at three equal positions at intervals, and connect the center of each imaging surface 39a of the imaging element 39 in the camera and the center of the imaging region 29a of the film 29. Each is arranged on a line. As a result, the detection optical systems 37a, 37b, and 37c are arranged so as to be shifted to the side closer to the Z-axis than the corresponding imaging elements 39, as shown in FIG. 9B. FIG. 9B is a side view of a surface including the photographing optical axis of the bright field image camera 35 and the principal ray of the scattered light 34b incident on the dark field image camera 36b. .

  With such an optical system, the image of the film 29 formed on each of the dark field image cameras 36a to 36c is substantially focused on the entire surface, and the image has almost no geometric distortion caused by the inclination of the optical system. There will be no image.

  Next, the configuration of the image processing unit 5B will be described with reference to FIG.

  The image processing unit 5B includes a bright field image acquisition unit 11B for capturing bright field image data input from the bright field image camera 35 and dark field image data from the dark field image cameras 36a to 36c. , And dark field image acquisition units 11a to 11c for capturing the images.

  The bright field image acquisition unit 11B provides the bright field image recording unit 21 and the dark field image acquisition units 11a-11c provide the dark field image recording units 12a-12c as digitized electronic image data. Each is to be recorded.

  The dark field images recorded in the dark field image recording units 12a to 12c are transferred to the wound image extraction processing units 14a to 14c, respectively, and the bright field images recorded in the bright field image recording unit 21 are It is transferred to all of the flaw image extraction processing units 14a to 14c. The flaw image extraction processing units 14a to 14c normalize the received dark field image and bright field image with respective average brightness values, and standardize the bright field image from each normalized dark field image. Processing for extracting a scratch image by subtracting each is performed, and these processing are substantially the same as those described in the second embodiment. However, in the third embodiment, since the bright field image and the dark field image are image data input from different cameras, the flaw image extraction processing units 14a to 14c will associate pixel positions between the cameras later. Then, the subtraction process between the standardized images is performed.

  Further, the other configuration of the flaw detection processing unit 13 and the configuration of the flaw correction processing unit 22 are the same as those in the second embodiment.

  Next, with reference to FIG. 12, the flow of processing by this scratch detection apparatus that also serves as a film digitizing apparatus will be described.

  FIG. 12A is a flowchart showing the system operation of the flaw detection apparatus.

  As shown in FIG. 12A, the operation of this flaw detection system is roughly classified into the following four processes.

  First, the calibration process between the cameras is performed for the four cameras, that is, the bright field image camera 35 and the dark field image cameras 36a, 36b, and 104c (step S20). By performing this calibration process, a flawed pixel correction table is created as will be described later. This process only needs to be performed once when this system is operated.

  The subsequent processing from step S21 to step S23 is processing that is repeatedly performed for each frame of the film 29.

  That is, first, the transport device (not shown) moves so that the center of the photographing region 29a that is the photographing frame of the film 29 is positioned on the Z axis. And the optical image currently recorded on the film 29 is taken in as a bright field image and a dark field image (step S21).

  Next, using the captured bright-field image (film original image) and the three dark-field images and the flaw pixel correction table created by the calibration process in step S20, a flaw pixel in the film original image is used. Scratch detection processing for detecting the position is performed, and scratch detection image data is created (step S22).

  Thereafter, if the scratch detection image data includes scratch information, a scratch correction process for correcting the scratch of the bright field image (film original image) is performed based on the information (step S23).

  Such a series of processing from film frame advancement to image capture and image processing is repeated until the shooting frames recorded on the film 29 are completed. However, it is also possible to perform film processing and image capture for each recorded image after performing film advance and image capture for all shot frames in advance. This is the same as in the second embodiment described above.

  Next, FIG. 12B is a flowchart showing details of the calibration process between the cameras in step S20.

  This calibration process accurately determines which pixel position of the image obtained by the bright field image camera 35 the pixel position determined to be a flaw based on the images of the dark field image cameras 36a to 36c. This is a process of creating a pixel correspondence map between cameras for associating with.

  That is, in each camera 35, 36a to 36c, magnification, image distortion, or the like between the imaging optical system 35a and each of the detection optical systems 37a to 37c, or between the detection optical systems 37a to 37c. There can be a difference.

  Further, errors in position and rotation position when the imaging optical system 35a or detection optical systems 37a to 37c are attached to the cameras 35 and 36a to 36c, or attachment positions and rotations of the cameras 35 and 36a to 36c themselves in the scratch detection device. There may also be a position error.

  Therefore, it is possible to correct these errors by performing this calibration process, that is, a process for obtaining a correction amount necessary for performing correction by measuring these error amounts in advance. It has become.

  When the calibration process is started, first, the center of the reference chart is set so as to be located on the Z axis (step S51). This reference chart is obtained by recording a reference image for mapping processing on a film or a quartz plate.

  Next, the images of the reference chart are taken by the cameras 35, 36a to 36c, and the camera coordinate values corresponding to the same point on the reference chart are obtained in the obtained images (step S52).

  Based on the obtained camera coordinate values, a map of pixel positions of the bright field image camera 35 corresponding to all pixel positions of the dark field image cameras 36a to 36c is obtained for each dark field image camera 36a to 36c. (Step S53).

  After that, a flaw pixel correction table as a mapping table is created for all the dark field image cameras 36a to 36c (step S54), and this calibration process is terminated.

  Next, FIG. 12C is a flowchart showing details of the image capturing process in step S21.

  When this process is started, as described above, the center of the shooting area 29a, which is the shooting frame of the film 29, is moved so as to be positioned on the Z axis, and then the illumination device 31 is turned on, so that a substantially white color is obtained. The illumination light 32 is irradiated toward the imaging region 29a (step S61).

  The imaged area 29a thus illuminated is captured simultaneously by the bright-field image camera 35 and the dark-field image cameras 36a to 36c, so that an image is captured and transferred to the image processing unit 5B. The image processing unit 5B records the image data transferred by the image acquisition units 11B and 11a to 11c as digitized electronic image data in the image recording units 21 and 12a to 12c described above. (Step S62), the image capturing process is terminated.

  Further, the scratch detection process in step S22 is substantially the same as that described with reference to FIG. 6C in the second embodiment.

  However, when subtracting the standardized bright field image data from the standardized dark field image data in step S43, each standardized dark field image is obtained using the flaw pixel correction table created in the calibration process of step S20. The correction is performed so that the pixel position of the data corresponds to the pixel position of the standardized bright field image, and then the luminance value is subtracted for each corresponding pixel. Thereby, the above-mentioned individual difference for each camera, that is, the shift of the pixel position of each image data can be corrected, and an accurate subtraction process can be performed.

  Further, the scratch correction process in step S23 is the same as that in the second embodiment.

  In the above description, the dark field image is captured from three directions, but the present invention is not limited to this, and the image may be captured from two, or four or more directions. If the direction in which the dark field image is captured is increased, it is possible to obtain a scratch image with fewer undetected scratches.

  As described in Example 2 above, the film is likely to have many scratches in the running direction, but such scratches may be fine scratches with low contrast that do not affect the original image. Many. Therefore, there is a tendency that the flaws in this direction are detected excessively, and a lot of time may be required for the flaw determination process. Therefore, the dark field image camera in which the line of the principal ray incident on the dark field image camera projected on the film surface does not become a direction orthogonal to the film traveling direction (more widely, a direction independent of the film traveling direction). A dark field in which the line is a direction dependent on the film traveling direction (same direction as or opposite to the film traveling direction) by taking the arrangement (ie, for example, a line in which the principal ray of the dark field image camera is projected onto the film surface) By taking an image camera arrangement), the processing speed of the flaw detection can be improved.

  On the other hand, the processing may be slow, but if it is desired to prevent the scratch from being undetected, a line obtained by projecting the principal ray to the dark field image camera onto the film surface If the dark-field image camera is arranged so as to be in a direction orthogonal to the traveling direction (more broadly, a direction independent of the film traveling direction), the scratch in the traveling direction can be detected with high accuracy, Scratch detection performance can be improved.

  Further, in the third embodiment, the flaw detection is performed using both the bright field image and the dark field image, but in the same manner as described with reference to FIG. 2 in the above-described first embodiment. It is also possible to perform a flaw detection process using only a dark field image. Specifically, the flaw detection is performed based on the flaw detection program as shown in FIG. 3 using the dark field image information obtained by imaging with the dark field image cameras 36a to 36c shown in FIG. May be.

  According to the third embodiment, one illuminating device that emits substantially white illumination light, one bright-field image camera that images the illuminated subject, and one or more (this third embodiment). By using three dark field image cameras, the same effects as those of the first and second embodiments can be obtained.

  Since a bright-field image and a dark-field image can be acquired simultaneously by a single imaging operation, the image can be captured at high speed, and a film digitizing apparatus or the like that is preferably equipped with a high-speed scratch detection function or the like is desirable. Can be applied well.

  In each of the above-described embodiments, a method using transmitted light has been described as an example. However, the present invention is not limited to this, and is also applicable to a method for detecting scratches using reflected light. Is possible. However, in this case, the reflected light on the front and back of the subject must be photographed separately, and depending on the shape of the subject, photographing from a different direction is required, which complicates the system. This increases the time required for processing. Furthermore, since reflected light is easier to pick up specularly reflected light from the cross section of the scratch than transmitted light, there is a tendency to detect excessive scratches. Therefore, when the subject is a transmissive object such as a film, it is considered that the method using transmitted light is more reasonable. As described above, it is desirable to appropriately select whether to use a system using transmitted light or a system using reflected light depending on the subject.

  At this time, a system that can use both transmitted light and reflected light may be constructed to enable detection according to various types of subjects.

  The subject is not limited to a transparent film or a film on which an image is recorded, and may be a subject having light transmissivity such as a glass substrate or an optical lens, a metal part, a painted part, a silicon wafer It is also possible to use various types of specimens that do not have optical transparency.

  Since some objects are colored, it is preferable to select an appropriate wavelength range for the light to be irradiated according to the color of the object. In addition, it is preferable to select the wavelength range of the light to be irradiated according to the size of the scratch or foreign object to be inspected.

  In addition, the detection target is not limited to scratches on the subject, foreign matter such as dust attached to the subject, for example, in the case of a subject configured by joining a plurality of parts, It can also be used to detect misalignment or the like of the joint portion. In this way, not only scratches and foreign objects but also objects that cause optical shadows or images when illuminated with illumination light can be widely detected.

  In the above description, the processing shown in FIGS. 3, 6, 12, and the like has been described as an operation in the flaw detection apparatus, but may be executed as a flaw detection program in a computer.

  In addition, this invention is not limited to the Example mentioned above, Of course, a various deformation | transformation and application are possible within the range which does not deviate from the main point of invention.

  INDUSTRIAL APPLICABILITY The present invention can be used for inspecting scratches, foreign matter, and the like on a subject, and is particularly effective for inspecting direction-dependent scratches, foreign matter, etc. on a subject having permeability such as a film. Can be used.

BRIEF DESCRIPTION OF THE DRAWINGS (A) A perspective view and (B) top view which show the outline of a structure of the wound detection apparatus in Example 1 of this invention. FIG. 2 is a block diagram mainly showing an outline of a configuration of an image processing unit in the first embodiment. 7 is a flowchart illustrating a flaw detection process performed by the image processing unit according to the first embodiment. The perspective view which shows the outline of a structure of the wound detection apparatus in Example 2 of this invention. FIG. 6 is a block diagram mainly showing an outline of a configuration of an image processing unit in the second embodiment. The flowchart which shows each the processing content by the flaw detection apparatus of the said Example 2. FIG. In the said Example 2, the figure which shows the dark field image obtained by each image | photographing a film with the illumination from three directions, and the process image obtained by processing these dark field images in two ways. The perspective view which shows the outline of a structure of the wound detection apparatus in Example 3 of this invention from the bottom face side. The bottom view and side view which show arrangement | positioning with the film surface, lens, and imaging surface in the flaw detection apparatus of the said Example 3. FIG. In the said Example 3, the figure which shows the example of the positional relationship of a film surface, a lens, and an imaging surface. FIG. 10 is a block diagram mainly showing an outline of a configuration of an image processing unit in the third embodiment. The flowchart which shows each the processing content by the flaw detection apparatus of the said Example 3. FIG.

Explanation of symbols

1a ... R dark field illumination device (illumination means, light source)
1b ... G dark field illumination device (illumination means, light source)
1c ... B dark field illumination device (illumination means, light source)
4 ... Transmission image camera (imaging means, color camera)
5, 5A, 5B ... image processing unit 9, 29 ... film (subject)
DESCRIPTION OF SYMBOLS 11 ... Dark field image acquisition part 11A ... Bright / dark field image acquisition part 11B ... Bright field image acquisition part 11a, 11b, 11c ... Dark field image acquisition part 12a, 12b, 12c ... Dark field image recording part (dark field image recording) means)
13. Scratch detection processing unit 14a, 14b, 14c ... Scratch image extraction processing unit (scratch image extraction processing means)
15. Scratch image integration processing unit (scratch image integration processing means)
16: Scratch determination unit (scratch determination means)
21 ... Bright field image recording unit (bright field image recording means)
DESCRIPTION OF SYMBOLS 22 ... Scratch correction process part 24 ... Acquisition timing generator 25 ... Light source switching machine 26 ... Bright field illumination device (2nd illumination means)
31 ... Illumination device (illumination means)
35 ... Bright field image camera 35a ... Imaging optical system (second optical means)
36a, 36b, 36c ... dark field image cameras 37a, 37b, 37c ... detection optical system (optical means)
38 ... Image sensor (second image sensor)
39: Image sensor
Agent Patent Attorney Susumu Ito

Claims (24)

A wound detection device for detecting foreign matter or scratches on a subject,
Illuminating means for irradiating a subject with illumination light of a plurality of types of wavelength ranges;
Imaging means for imaging the subject illuminated by the illumination means as a dark field image;
A flaw detection device comprising:   The illumination means irradiates illumination light of a plurality of types of wavelength regions such that an overlap region where the wavelength regions overlap each other is a sufficiently narrow region with respect to each wavelength region. Wound detection device according to.   The plurality of types of wavelength ranges include at least two wavelength ranges among a wavelength range expressing red, a wavelength range expressing green, and a wavelength range expressing blue. Scratch detection device. The illuminating means is configured by arranging a plurality of light sources corresponding to the plurality of types of wavelength ranges so as to irradiate the subject with illumination light from different directions,
2. The imaging device according to claim 1, wherein the imaging unit is configured to simultaneously image a subject illuminated with illumination light of a plurality of types of wavelength ranges simultaneously irradiated from the plurality of light sources. Wound detection device according to. The subject is conveyed relative to the imaging means in one direction along a plane whose normal is the imaging direction of the imaging means,
5. The flaw detection according to claim 4, wherein at least one of the plurality of light sources has a direction in which a direction in which the subject is illuminated is projected onto the plane is a direction independent of the transport direction. apparatus. The plurality of types of wavelength ranges include at least a wavelength range expressing red,
The light source for irradiating illumination light in a wavelength region expressing red is a light source in which a direction in which a direction of illuminating a subject is projected onto the plane is a direction independent of the transport direction. 5. The wound detection apparatus according to 5. The subject is conveyed relative to the imaging means in one direction along a plane whose normal is the imaging direction of the imaging means,
5. The flaw detection according to claim 4, wherein at least one of the plurality of light sources has a direction in which a direction in which the subject is illuminated is projected onto the plane is a direction dependent on the transport direction. apparatus. The plurality of types of wavelength ranges include at least a wavelength range expressing red,
The light source for irradiating illumination light in a wavelength range expressing red is a light source in which a direction obtained by projecting a direction of illuminating a subject onto the plane is a direction dependent on the transport direction. 8. The wound detection apparatus according to 7. A second illumination means for irradiating the subject with white illumination light;
The wound detection apparatus according to claim 4, wherein the imaging unit is configured to capture a subject illuminated by the second illumination unit as a bright field image.   The wound detection apparatus according to claim 4, wherein the imaging unit includes a color camera capable of imaging a subject illuminated with illumination light in the plurality of types of wavelength ranges.   The imaging device includes a color camera capable of imaging the subject illuminated with the illumination light of the plurality of types of wavelength regions and the subject illuminated with the white illumination light. 10. The wound detection apparatus according to 9. Dark field image recording means for extracting and recording dark field images for each of the plurality of types of wavelength regions from image data obtained by imaging by the imaging means;
The brightness average value of the dark field image recorded in the dark field image recording means is calculated, the dark field image is normalized by the brightness average value, and the pixel value is calculated based on the image data of the normalized dark field image. Wound image extraction processing means for extracting local variation data as wound image data;
Flaw image integration processing means for extracting the maximum value for each pixel based on the flaw image data for each of the plurality of types of wavelength ranges extracted by the flaw image extraction processing means and generating one integrated flaw image data When,
Based on the flaw image data generated by the flaw image integration processing means, the pixel value is compared with a predetermined threshold value, and when the pixel value is equal to or greater than the predetermined threshold value, the flaw is determined as a foreign object or a flaw on the subject. A determination means;
The flaw detection device according to claim 4, further comprising: Dark field image recording means for extracting and recording dark field images for each of the plurality of types of wavelength regions from image data obtained by imaging the subject illuminated by the illumination means by the imaging means;
Bright field image recording means for recording image data obtained by the imaging means imaging the subject illuminated by the second illumination means as a bright field image;
The brightness average value of the dark field image recorded in the dark field image recording means is calculated, the dark field image is normalized by the brightness average value to generate normalized dark field image data, and the bright field image The brightness average value of the bright field image recorded in the recording means is calculated, the bright field image is normalized by the brightness average value, and the standardized bright field image data is generated. A difference image data obtained by subtracting the normalized bright field image data, and based on the difference image data, a flaw image extraction processing unit that extracts local variation data of pixel values as flaw image data;
Flaw image integration processing means for generating a single integrated flaw image data by extracting the maximum value for each pixel based on the flaw image data for each of the plurality of types of wavelength ranges extracted by the flaw image extraction processing means. When,
Based on the flaw image data generated by the flaw image integration processing means, the pixel value is compared with a predetermined threshold value, and when the pixel value is equal to or greater than the predetermined threshold value, the flaw is determined to be a foreign matter or a flaw on the subject. A determination means;
The flaw detection device according to claim 9, further comprising: A wound detection device for detecting foreign matter or scratches on a subject,
An illumination unit configured to irradiate illumination light from a direction substantially orthogonal to the surface of the subject, a light source that emits substantially white illumination light that is visible light;
At least one dark field image camera configured to capture a subject illuminated with illumination light from the light source as a dark field image;
A flaw detection device comprising:   15. The flaw detection apparatus according to claim 14, further comprising a bright field image camera configured to capture a subject illuminated by illumination light from the light source as a bright field image. The subject is conveyed relative to the illumination means in one direction along a plane whose normal is the illumination direction by the illumination means,
The at least one of the dark-field image cameras has a direction in which a direction in which a subject is imaged is projected on the plane is a direction independent of the transport direction. Scratch detection device. The subject is conveyed relative to the illumination means in one direction along a plane whose normal is the illumination direction by the illumination means,
The at least one of the dark-field image cameras has a direction dependent on the transport direction as a direction in which a direction in which a subject is imaged is projected onto the plane. Scratch detection device. A plurality of the dark field image cameras are provided, and the plurality of dark field image cameras are configured to have different directions for imaging the subject,
Dark field image recording means for recording a plurality of dark field image data for different imaging directions obtained by imaging by the plurality of dark field image cameras;
Bright field image recording means for recording bright field image data obtained by imaging by the bright field image camera;
The brightness average value of the dark field image recorded in the dark field image recording means is calculated, the dark field image is normalized by the brightness average value to generate normalized dark field image data, and the bright field image The brightness average value of the bright field image recorded in the recording unit is calculated, the bright field image is normalized by the brightness average value, and the standardized bright field image data is generated. A difference image data obtained by subtracting the normalized bright field image data, and based on the difference image data, a flaw image extraction processing unit that extracts local variation data of pixel values as flaw image data;
A wound image integration processing unit that extracts a maximum value for each pixel based on the plurality of wound image data extracted by the wound image extraction processing unit and generates one integrated wound image data;
Based on the flaw image data generated by the flaw image integration processing means, the pixel value is compared with a predetermined threshold value, and when the pixel value is equal to or greater than the predetermined threshold value, the flaw is determined as a foreign object or a flaw on the subject. A determination means;
The flaw detection device according to claim 15, further comprising: The subject is a film-like object having a planar surface,
The dark field image camera includes an optical unit for forming an optical image of the subject, an imaging element that converts an optical image formed on the imaging surface by the optical unit into an electrical signal, and Comprising:
The imaging element is arranged so that its imaging surface is substantially parallel to the planar surface of the subject,
The optical means acts so that the imaging surface and the planar surface of the subject have a conjugate relationship,
Of the light rays from the planar surface of the subject to the imaging surface, the principal ray passing through the center of the imaging surface is configured to intersect with the optical axis of the optical means. The flaw detection device according to claim 14. The subject is a film-like object having a planar surface,
The dark field image camera includes an optical unit for forming an optical image of the subject, an imaging element that converts an optical image formed on the imaging surface by the optical unit into an electrical signal, and Comprising:
The imaging element is arranged so that its imaging surface is substantially parallel to the planar surface of the subject,
The optical means acts so that the imaging surface and the planar surface of the subject have a conjugate relationship,
Of the light rays from the planar surface of the subject to the imaging surface, the principal ray passing through the center of the imaging surface is configured to intersect with the optical axis of the optical means. The flaw detection device according to claim 15. The bright field image camera is disposed at a position where the transmitted light of the subject illuminated by the illumination light from the light source can be captured as a bright field image, and forms an optical image of the subject. And a second imaging device for converting an optical image formed on the imaging surface by the second optical means into an electrical signal. Yes,
The second imaging element is arranged so that its imaging surface is substantially parallel to the planar surface of the subject,
The second optical means acts so that the imaging surface of the second imaging element and the planar surface of the subject have a conjugate relationship,
Of the light rays from the planar surface of the subject to the imaging surface of the second image sensor, the principal ray passing through the center of the imaging surface is substantially parallel to the optical axis of the second optical means. The flaw detection apparatus according to claim 20, wherein the flaw detection apparatus is configured as described above. A flaw detection program for detecting a foreign object or a flaw on a subject,
On the computer,
A first procedure for reading a dark field image from a plurality of directions of a subject illuminated by illumination light from a light source;
A second procedure for recognizing a portion having a high luminance value among the plurality of dark field images read in the first procedure as a foreign matter or a flaw on the subject;
Scratch detection program for running. The second procedure is as follows:
A third procedure for normalizing the plurality of dark field images read in the first procedure by the average luminance value of each dark field image;
A fourth procedure for extracting a top hat value, which is a high-frequency peak, from a local change amount in the vicinity of the pixel for each pixel of the plurality of images standardized in the third procedure;
A fifth procedure for selecting a maximum value among the top hat values of the plurality of images extracted in the fourth procedure and generating one image from the plurality of images using the selected maximum value;
A sixth procedure for comparing a top hat value in one image generated in the fifth procedure with a predetermined threshold and selecting a pixel portion such as a foreign object or a flaw;
The flaw detection program according to claim 22, comprising: The second procedure is as follows:
A seventh procedure for normalizing the plurality of dark field images read in the first procedure by the average luminance value of each dark field image;
An eighth procedure for reading a bright-field image of a subject illuminated by illumination light from a light source;
A ninth procedure for normalizing the bright field image read in the eighth procedure with the average luminance value of the bright field image;
A tenth procedure for subtracting the luminance value of the bright field image standardized in the ninth procedure from each dark field image standardized in the seventh procedure;
An eleventh procedure for extracting a top hat value that is a high-frequency peak from a local change amount in the vicinity of the pixel for each pixel of the plurality of subtraction results obtained in the tenth procedure;
A twelfth procedure for selecting a maximum value among the top hat values of the plurality of images extracted in the eleventh procedure and generating one image from the plurality of images using the selected maximum value;
A thirteenth procedure for selecting a pixel portion such as a foreign object or a flaw by comparing a top hat value in one image generated in the twelfth procedure with a predetermined threshold;
The flaw detection program according to claim 22, comprising: