Inspection method of fiber cross section

Abstract

To provide an inspection method of a fiber cross section for proper control of the cross section in cross section inspection of a fiber made of two or more different polymers.SOLUTION: An inspection method of a fiber cross section in cross section inspection of a fiber made of two or more different polymers includes mask processing to hide a part of respective images of a sample image and an inspection image for detection of the profile of the fiber and the profile that separates the polymers. Using an image of a specific part, the coincidence rate of the images is determined to estimate the profile of the fiber and the profile that separates the polymers.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a method for inspecting the cross section of a synthetic fiber made of two or more types of polymers, and more particularly to a method for detecting the contours of the fiber and the contours separating the polymers in a complex cross-sectional shape. [Background technology]

[0002] Synthetic fibers made from thermoplastic polymers such as polyester and polyamide have excellent mechanical properties and dimensional stability, and are therefore widely used in a variety of applications, from clothing to interior and vehicle interiors, industrial applications, etc. As the properties required of materials become more diverse, a variety of functional fibers have been developed.

[0003] For example, by creating a side-by-side structure in which two different types of polymers are bonded to each other, fibers with crimps that correspond to the difference in shrinkage rates between the polymers have been developed, and by laminating two or more types of polymers in multiple layers parallel to the fiber surface, fibers with high opacity, UV protection, heat insulation, and a soft texture have been developed.

[0004] To ensure these various functionalities, many image processing and sensor technologies have been proposed to inspect for defects such as fuzz, leading to the supply of high-quality, highly stretchable fibers and other added-value fibers.

[0005] However, in fibers with a cross-section where two or more different polymers are arranged in a complex manner, even a slight defect in polymer dispensing can result in fibers being washed out, which can have a significant impact on functionality and quality. For example, in fibers made of two or more different polymers, one of which is an easily soluble polymer, poor dispensing of the easily soluble polymer can prevent the other components from peeling off, resulting in the failure to exhibit functionality. In order to exhibit functionality, it is important to inspect all fibers to ensure that they have the desired fiber cross-section and to manage the fiber cross-section.

[0006] Furthermore, if the fiber is made of one type of polymer and has a round cross section, it is possible to distinguish the fiber from the difference in brightness between the background and the fiber, and the shape of each single yarn can be measured immediately. However, if the fiber contains two or more different polymers, it is difficult to determine the outline of the fiber and the outline separating the polymers from the difference in brightness, so visual inspection is required. This requires a huge amount of visual inspection, making complete measurement difficult and creating the risk of overlooking something due to operational errors.

[0007] To solve these problems, Patent Document 1 proposes an image analysis device and program for analyzing polished cross-sectional images of fiber-reinforced composite materials. Specifically, the proposed image analysis device and program can easily calculate the fiber area by estimating the center and radius of the fiber based on an energy function that utilizes the fact that the contours of the searched fibers are circular. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Japanese Patent Application Laid-Open No. 2015-45997 Summary of the Invention [Problem to be solved by the invention]

[0009] In fibers with a cross section where two or more different polymers are arranged in a complex manner, it is extremely important to judge the contours that make up the fiber and the contours that separate the polymers. In particular, in fibers that utilize the difference in shrinkage rates between polymers, if the contours are not formed correctly, functionality will not be expressed and this will result in a product defect. Furthermore, the more complex the cross section, the more contours that separate the polymers, making it more difficult to judge.

[0010] Patent Document 1 proposes an image analysis device equipped with a fiber region determination processing unit that estimates the center and radius of a fiber based on an energy function that takes advantage of the fact that the contour of the fiber being searched for is circular. However, for fibers with a cross section in which two or more different types of polymers are arranged in a complex manner, it is not possible to estimate the contours that make up the fiber and the contours that separate the polymers and determine the fiber region. In the image processing of Patent Document 1, a contour extraction process is performed using an image filter to extract the contours of voids, and then the contours of fibers are extracted using the image filter again.The center and radius of the fiber region are then estimated based on an energy function to determine the fiber region.The image filter of Patent Document 1 performs spatial filtering, and determines pixel values ​​of the output image using pixel values ​​within the region, including not only the corresponding pixel value of the input image but also the surrounding pixels. The image processing technique described in Patent Document 1 is appropriate for fibers with a circular cross section made of one type of polymer, but for fibers with a cross section made up of two or more different polymers arranged in a complex manner, it is not possible to distinguish between the contours of the fibers and the contours separating the polymers, making it impossible to determine the fiber region. The pixel values ​​corresponding to the contours of the fibers and the contours separating the polymers show the same color scheme, and the pixel values ​​of the complexly arranged polymers have a large influence, making it difficult to extract the contours appropriately.

[0011] The present invention provides a method for determining the shape of a fiber region in a cross-section of a fiber in which two or more different polymers are arranged in a complex manner, by estimating the contours that define the fiber and the contours that separate the polymers. [Means for solving the problem]

[0012] In order to solve the above problems, the present invention comprises the following configuration. (1) An inspection method for fiber cross sections made of two or more different polymers, using an apparatus including an image acquisition unit that inputs inspection image data and an image evaluation unit that calculates the difference between the pixel values ​​of a sample image and the pixel values ​​of an inspection image and calculates the match rate by using the sum of the absolute values ​​of the difference, A fiber cross-section inspection method in which, in order to detect the contours of the fibers and the contours separating the polymers, a masking process is performed to hide part of each of the specimen image and the test image, and the image of the specific part is used to calculate the image match rate and estimate the contours of the fibers and the contours separating the polymers. (2) The fiber cross-section inspection method according to (1) above, wherein areas other than the target fiber cross-section are masked in the inspection image. (3) A fiber cross-section inspection method according to (2) above, in which an area range corresponding to a region containing 50% or less of the length of a line segment extending from the center of the fiber toward the outline of the fiber is masked in the fiber cross-section of the specimen image and the inspection image. (4) A fiber cross-section inspection method according to (2) above, in which an area range corresponding to a region including a line segment extending from the fiber outline to the fiber center in the fiber cross-section of the specimen image and the inspection image is masked. [Effects of the Invention]

[0013] According to the fiber cross-section inspection method of the present invention, by performing appropriate masking processing on a fiber cross-section composed of two or more different polymers, it is possible to determine the contours of the fiber and the contours separating the polymers, thereby making it possible to properly inspect the fiber cross-section. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a schematic diagram of an inspection image of a cross section of a fiber made of two or more different polymers. [Figure 2] FIG. 2 is a schematic diagram of a sample image of the fiber to be inspected in FIG. 1. [Figure 3] 1A and 1B are schematic diagrams of an inspection image illustrating a first embodiment of the present invention. [Figure 4] FIG. 10 is a schematic diagram of an inspection image showing a second embodiment of the present invention. [Figure 5] FIG. 10 is a schematic diagram of an inspection image showing a third embodiment of the present invention. [Figure 6] FIG. 10 is a schematic diagram of a specimen image showing a third embodiment of the present invention. [Figure 7] FIG. 10 is a schematic diagram of an inspection image showing a fourth embodiment of the present invention. [Figure 8] FIG. 10 is a schematic diagram of a specimen image showing a fourth embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0015] The following description will be given with reference to the drawings. Figure 1 is a schematic diagram of an inspection image of a fiber cross section composed of two or more different polymers. The inspection image consists of the fiber cross section 1 to be inspected and a coating made of resin 2. For example, multiple fibers as the inspection sample are placed in a container, the surface of the fiber bundle is coated with resin 2, etc., and the hardened coating is cut to obtain the fiber cross section. The obtained fiber cross section is then photographed using an optical microscope to obtain an inspection image. It is preferable to photograph the fiber cross section to be inspected as large as possible, so that the inspection object occupies more than 30% of the entire image. The resin 2 used for the coating can generate noise 3, such as bubbles when the resin hardens or scratches when cutting, so it is preferable to minimize this noise.

[0016] In this embodiment, the matching rate and position matching rate with the specimen image are calculated using an image processing method described later to determine whether the inspection image is the target fiber cross section.

[0017] Here, the match rate is calculated by calculating the absolute value of the difference between the pixel values ​​of an arbitrary section of the specimen image and an arbitrary section of the test image, and dividing the sum by the number of pixels. A threshold value is set for the match rate to determine whether the image is the fiber cross-section of the test subject. It is preferable to use the same number of pixels for the specimen image and the test image, and to use images with as high a resolution as possible. Regarding the specific pixel count, for a typical clothing fiber thickness of 20 to 200 dtex, it is preferable to magnify the target fiber cross-section with a microscope by approximately 200 times and use image data acquired with a vertical pixel count of 1440 x horizontal pixel count of 1080, or 1.56 million pixels or more. These pixel counts may be adjusted appropriately depending on the thickness of the fiber cross-section and the imaging equipment used; they are not limited to these values.

[0018] Furthermore, the user inputs the fiber contours 4 and the contours 5 separating the polymers into the specimen image in advance, and the inspection method described below is used to check the degree of agreement between the fiber contours 4 and the contours 5 separating the polymers, and then the inspection is performed.

[0019] The position matching rate is calculated by comparing the outline of the fiber set in the specimen image with the inspection image, calculating the absolute value of the difference between the pixel values ​​of the outline of the fiber in the specimen image and the pixel values ​​of an arbitrary section of the inspection image, and dividing the sum by the number of pixels. The section with the highest position matching rate obtained by this calculation method is determined as the position of the fiber cross section, and the fiber region is then determined.

[0020] The inspection procedure is as follows: First, the minimum distance from the center of the fiber to the outline of the fiber is measured in the specimen image, and then the specimen image is searched from any point where the specimen image is included in the inspection image.

[0021] For example, if the minimum distance from the center of the fiber to the fiber contour in the specimen image is 100 pixels, the search begins at a point 100 pixels vertically and 100 pixels horizontally in the inspection image. The absolute value of the difference between the pixel values ​​of the fiber contour in the specimen image and the pixel values ​​of an arbitrary section centered on the search start point is calculated. If the sum of the absolute values ​​of these differences exceeds a threshold, the arbitrary section and the specimen image are determined to not match, and the system moves on to the next section. The system searches for positional match rates, including the rotational direction, and the section with the highest positional match rate is determined to be the location of the fiber cross section 1 to be inspected. Then, at the point estimated as the fiber cross section 1, a search is performed again in the rotational direction on the specimen image, where the contour 5 separating the polymer is set. The highest match rate is extracted, and a user-specified threshold is used to determine whether a normal fiber cross section 1 has been formed.

[0022] The inspection image shown in Figure 1 is susceptible to errors in the measurement of the matching rate due to disturbances other than the fiber cross section. Because the fiber contour 4 may not be uniform due to the influence of cutting the fiber-containing coating, simply determining the position of the fiber cross section can lead to a decrease in the accuracy of the matching rate. Even if the fiber cross section is normal, the matching rate may be judged to be low, which could lead to false detection of an abnormal fiber cross section.

[0023] Therefore, by satisfying the embodiment of the present invention, normal fiber cross sections can be extracted with a high degree of accuracy and accurate judgment can be made. FIG. 3 is a schematic diagram of an inspection image showing a first embodiment of the present invention. The inspection image of FIG. 1 is affected by noise 3 generated in the resin 2, making it difficult to determine the same fiber cross-section as the specimen image of FIG. 2. Specifically, noise 3, such as scratches generated when cutting a fiber-containing coating, may overlap with the fibers, potentially leading to the risk of misjudging the scratches (noise) generated in the resin 2 as the position of the contour. Therefore, in FIG. 3, masking 7, which hides part of the inspection image, allows for accurate determination of the fiber cross-section alone. While the location where masking 7 is performed is not limited, masking the resin with noise and part of the fiber cross-section allows for accurate position matching and prevents erroneous determination of the matching rate. The image processing method will be described later.

[0024] 4 is a schematic diagram of an inspection image showing a second embodiment of the present invention. The image processing method will be described later, but this is an image in which only fiber cross sections in which the entire fiber cross sections are shown on the image are inspected, and the rest of the image is masked. By masking the rest of the fiber cross sections other than the fiber cross sections to be inspected, the inspection target can be made clearer. In the image, missing fiber cross sections can be excluded from the inspection target.

[0025] 5 is a schematic diagram of an inspection image showing a third embodiment of the present invention. The image processing method will be described later, but this image is an inspection image showing the second embodiment, in which an area range equivalent to a region including a length of 50% or less of the line segment extending from the center 6 of the fiber toward the outline 4 of the fiber is masked.

[0026] By masking the center of the fiber cross section to be inspected, the match rate is judged without inspecting the center of the fiber cross section. By specifying the area to be judged, noise 3 that covers the fiber cross section can be excluded, reducing false positives. Furthermore, in fiber cross sections where multiple types of polymers are arranged in a complex manner and there are many contours 5 that separate the polymers, the sharpness of the fiber-containing coating when cutting can slightly shift the position of the contours 5 that separate the polymers, erroneously reducing the match rate of normal fiber cross sections and increasing false positives. Furthermore, in cross-sectional fibers such as hollow fibers where the influence of the fiber center is small, masking the center can reduce the risk of false positives.

[0027] FIG. 6 is a schematic diagram of a specimen image showing a third embodiment of the present invention. This is used to determine the match rate in the schematic diagram of the inspection image showing the third embodiment described above. As with the inspection image, the specimen image is masked to an area corresponding to a region containing 50% or less of the length of the line segment extending from the fiber center 6 toward the fiber outline 4. The range can be set to any value, but the masking is performed at the same rate as the masking 7 performed on the inspection image. Performing masking at the same rate prevents an erroneous decrease in the match rate, allows the position of the fiber cross section 1 in the inspection image to be accurately detected, and enables abnormality determination based on the match rate. This is effective for products with complex shapes in the center, such as hollow fibers.

[0028] FIG. 7 is a schematic diagram of an inspection image showing a fourth embodiment of the present invention. While the image processing method will be described later, this image is an image obtained by masking the inspection image showing the second embodiment of FIG. 4 by an area corresponding to a region including 50% or less of the length of the line segment extending from the fiber outline 4 toward the fiber center 6. By masking the outer periphery of the fiber cross section to be inspected, the outer periphery can be excluded from the judgment target and the match rate can be determined. Increasing the number of areas excluded from judgment allows for judgment to be made while excluding noise 3 on the fiber, reducing the number of false positives in which normal cross-sectional fibers are judged as abnormal. Furthermore, by placing an easily soluble polymer in the outermost layer and performing alkali weight reduction treatment, which is effective for fiber cross sections with complex cross-sectional shapes in the inner layer, unnecessary easily soluble polymer shapes can be excluded and judgment can be made. The masking range is not limited to 50% or less and can be set arbitrarily depending on the fiber cross-sectional shape.

[0029] FIG. 8 is a schematic diagram of a specimen image showing a fourth embodiment of the present invention. This is used to determine the match rate in the schematic diagram of the inspection image showing the fourth embodiment described above. As with the inspection image, the specimen image is masked to an area corresponding to a region containing 50% or less of the length of the line segment extending from the fiber outline 4 toward the fiber center 6. The range can be set to any value, but the masking is performed at the same rate as the masking performed on the inspection image. By masking at the same rate, an erroneous decrease in the match rate can be prevented, the position of the fiber cross section 1 in the inspection image can be accurately detected, and an abnormality can be determined from the match rate. This is effective for fibers with an easily soluble polymer arranged in the outermost layer, and the easily soluble polymer arranged in the outermost layer can be excluded to enable determination.

[0030] The schematic diagrams of fiber cross sections shown in Figures 1 to 8 show a circular cross section composed of two or more different polymers, but the fiber cross section shape is not limited to a circular cross section and can also be applied to flat or irregular cross sections. As shown in the image processing method described below, the center of the object is obtained from the design shape of the fiber cross section, but for an elliptical cross section, it can also be calculated from the focal position. The outline of the fiber and the center of the fiber can be determined according to a known fiber cross section design. The fiber center can be replaced with the center of gravity of the fiber cross section, but it is preferable that it does not lie outside the fiber cross section, and any position can be set according to the design shape of the cross section.

[0031] Next, an image processing method according to an embodiment of the present invention will be described. Although the image processing of the specimen image is not limited, first, when photographing the specimen image, the center of the object is captured and photographed to determine the center of the object.

[0032] In the case of a round shape, the distance from the fiber center estimated according to the fineness is set as an input value, and the image is cut out.

[0033] The cut-out image is subjected to mask processing to obtain a circular specimen image. For example, for a side-by-side fiber cross section, the distance from the center of the object to the long and short sides is entered, and the cutout is performed according to the outline of the fiber cross section design value.

[0034] Regardless of the fiber cross-sectional shape, the center position of the object is arbitrarily set when photographing, and the designed shape of the fiber cross-section is overlaid, and a sample image is obtained by performing cutout and mask processing.

[0035] The cutting method and mask processing method are not particularly limited, and known methods may be used.

[0036] Furthermore, for the specimen images showing the third and fourth embodiments, the center of the fiber is determined by the position at the time of photographing, and the contours of the fiber are determined from the fiber cross-section design values, so the distance connecting the center position of the fiber and the contours of the fiber can be calculated, and the contours to be masked can be determined.

[0037] Next, an image processing method for an inspection image according to the first embodiment will be described. In the image processing of the inspection image, first, the position to be masked is determined using the position matching rate obtained by searching the specimen image at the start of the inspection. Specifically, when starting a specimen image search using the fiber contours in an inspection image, the absolute difference in pixel values ​​between an arbitrary section and the specimen image is first calculated. Next, if the sum of the absolute values ​​of these differences exceeds a threshold, the arbitrary section is determined to not contain a specimen image, and the system moves on to the next section. In the section that does not contain a specimen image, a user-specified range is masked. This image processing removes noise other than the fiber cross section of the inspection target, preventing erroneous judgments.

[0038] Next, the image processing method for an inspection image according to the second embodiment uses the contours of the fibers to determine the position to be masked, excluding the section with the highest position matching rate. Specifically, the sample image is first searched for in the inspection image using the outline of the fiber. Next, the section with the highest positional match rate is determined to be the location of the target fiber cross section. Masking is then performed to remove the location of this fiber cross section. This image processing removes everything other than the fiber cross section being inspected, preventing erroneous judgments.

[0039] Next, the image processing method for an inspection image according to the third embodiment uses the inspection image processing according to the second embodiment to perform mask processing. Specifically, the fiber contours are used to determine the location of the target fiber cross section in the section with the highest positional match rate, and then a mask process is performed on an arbitrary range set in advance in the sample image from the center of the fiber to the fiber contour. By increasing the number of areas excluded from judgment, noise at the center of the fiber can be excluded and judgment can be made, reducing false positives.

[0040] Next, the image processing method for an inspection image according to the fourth embodiment uses the inspection image processing according to the second embodiment to perform mask processing. Specifically, the fiber contour is used to determine the location of the target fiber cross section in the section with the highest positional match rate, and then a mask process is performed on an arbitrary range from the fiber contour set in advance in the sample image to the center of the fiber. This is effective for fibers with easily soluble polymers arranged on the periphery, and makes it possible to make a judgment by excluding the range of the easily soluble polymer. [Example]

[0041] The present invention will be described in detail below with reference to examples. Example 1 Circular cross-section fibers composed of three different polymers were placed in an Eiko Sangyo Co., Ltd. automatic cross-section sampling device (model AYSM-T) to prepare the samples. Next, using a Keyence Corporation microscope (VHX-8000), the cross-section of the target fiber sample was magnified approximately 200 times and photographed at 1440 vertical and 1080 horizontal pixels (approximately 1.56 million pixels). Test and specimen images were then created at 400 vertical and 400 horizontal pixels, matching the size of the fiber cross-section. The captured images were processed as described above using OpenCV, an open-source library that integrates image and video processing functions. Object detection in the images was performed using OpenCV template matching. The position match rate was calculated from the sum of the pixel value differences between the fiber outlines in the specimen image and the test image. Areas exceeding 70% were determined to be fiber locations.

[0042] Next, the area where the position match rate was 5% or less was masked based on the sum of the differences between the outlines of the fibers in the specimen image and the pixel values ​​of the test image. After that, the match rate was calculated again for the area determined to be the fiber position, including the outlines separating the polymer, and if it exceeded 80%, it was determined to be a normal cross section.

[0043] When normal cross-sectional images and cross-sectional images containing abnormal areas were judged, the match rate for normal cross-sectional images was 83%, and the match rate for cross-sectional images containing abnormal areas was 71%. Judgment was possible at a match rate threshold of 80%.

[0044] (Comparative Example 1) The inspection was carried out in the same manner as in Example 1, except that the judgment was carried out using the unprocessed inspection image that was not subjected to masking. As a result of judging the normal cross-sectional image and the cross-sectional image including the abnormal part, the match rate for the normal cross-sectional image was 73%, and the match rate for the cross-sectional image including the abnormal part was 71%. The match rate for the normal cross-sectional image decreased, resulting in a false detection.

[0045] Example 2 The inspection was carried out in the same manner as in Example 1, except that areas other than the fiber cross section to be inspected were masked. Images of normal cross sections and images of cross sections containing abnormal areas were judged, and the match rate for images of normal cross sections was 86%, while the match rate for images of cross sections containing abnormal areas was 70%. Judgment was possible at a match rate threshold of 80%.

[0046] Example 3 Masking was performed on an area range corresponding to a region containing 50% or less of the length of the line segment extending from the center of the fiber to the fiber outline in the inspection image. Similarly, the specimen image was also masked on an area range corresponding to a region containing 50% or less of the length of the line segment extending from the center of the fiber to the fiber outline, but the inspection was performed in the same manner as in Example 2. Images of normal cross sections and images of cross sections containing abnormal areas were judged, and the match rate for normal cross sections was 95%, while the match rate for cross sections containing abnormal areas was 47%. A match rate threshold of 80% was the boundary for judgment.

[0047] Example 4 Masking was performed on an area range corresponding to a region containing 50% or less of the length of the line segment extending from the fiber outline toward the fiber center in the inspection image. Similarly, the specimen image was also masked on an area range corresponding to a region containing 50% or less of the length of the line segment extending from the fiber outline toward the fiber center, but the inspection was performed in the same manner as in Example 2. Images of normal cross sections and images of cross sections containing abnormal areas were judged, and the match rate for normal cross sections was 91%, while the match rate for cross sections containing abnormal areas was 57%. A match rate threshold of 80% was the boundary for making a judgment.

[0048] [Table 1]

[0049] Example 5 The test was carried out in the same manner as in Example 1, except that the sample was prepared using a side-by-side fiber consisting of two different polymers bonded to each other and configured with a flat, pot-shaped cross section. The center of the fiber was set as the point where the long side connecting the outlines of the fibers furthest from the paired polymers intersected perpendicularly with the short side where the paired polymers overlapped. Images of normal cross sections and images of cross sections containing abnormal areas were judged, and the match rate for normal cross section images was 87%, while the match rate for cross section images containing abnormal areas was 72%. A match rate threshold of 80% was used for judgment.

[0050] (Comparative Example 2) The inspection was carried out in the same manner as in Example 5, except that the judgment was carried out using the unprocessed inspection image without masking. As a result of judging the normal cross-sectional image and the cross-sectional image including the abnormal part, the match rate for the normal cross-sectional image was 79%, and the match rate for the cross-sectional image including the abnormal part was 73%. The match rate for the normal cross-sectional image decreased, resulting in a false detection.

[0051] Example 6 The inspection was carried out in the same manner as in Example 5, except that areas other than the fiber cross section to be inspected were masked. Images of normal cross sections and images of cross sections containing abnormal areas were judged, and the match rate for images of normal cross sections was 91%, while the match rate for images of cross sections containing abnormal areas was 65%. Judgment was possible at a match rate threshold of 80%.

[0052] Example 7 Masking was performed on an area corresponding to a region containing 30% or less of the length of the line segment extending from the center of the fiber toward the fiber's outline in the inspection image. In masking, the center of the fiber was determined as described above by using the point where the long and short sides intersect perpendicularly. The inspection was performed in the same manner as in Example 6, except that multiple straight lines were extended radially from the center of the fiber, and masking was performed on an area corresponding to a region containing 30% or less of the length of the line segment extending from the center of the fiber toward the fiber's outline, where the line segment extends from the center of the fiber to intersect with the fiber's outline. The specimen images were also subjected to the same processing. Images of normal cross sections and images of cross sections containing abnormal areas were evaluated. The match rate for normal cross sections was 92%, while the match rate for cross sections containing abnormal areas was 47%. Evaluation was possible at a match rate threshold of 80%.

[0053] Example 8 Masking was performed on an area corresponding to a region containing 30% or less of the length of the line segment extending from the fiber contour toward the fiber center in the inspection image. In masking, the center of the fiber was determined as described above by using the point where the long and short sides intersect perpendicularly. The inspection was performed in the same manner as in Example 6, except that multiple straight lines were extended radially from the center of the fiber, and masking was performed on an area corresponding to a region containing 30% or less of the length of the line segment extending from the fiber center to the point where it intersects with the fiber contour. Images of normal cross sections and images of cross sections containing abnormal areas were evaluated, and the match rate for normal cross sections was 93%, while the match rate for cross sections containing abnormal areas was 48%. Evaluation was possible at a match rate threshold of 80%.

[0054] [Table 2] [Explanation of symbols]

[0055] 1: Fiber cross section 2: Resin 3: Noise 4: Fibrous contours 5: Contours separating polymers 6: Fiber center 7: Masked area

Claims

1. An inspection method for a fiber cross section made of two or more different polymers, using an apparatus including an image acquisition unit that inputs inspection image data and an image evaluation unit that calculates the difference between the pixel values ​​of a sample image and the pixel values ​​of an inspection image and calculates a match rate by using the sum of the absolute values ​​of the difference, A fiber cross-section inspection method in which, in order to detect the contours of the fibers and the contours separating the polymers, a masking process is performed to hide part of each of the specimen image and the test image, and the image of the specific part is used to calculate the image match rate and estimate the contours of the fibers and the contours separating the polymers.

2. 2. The fiber cross-section inspection method according to claim 1, wherein the inspection image is masked to include areas other than the fiber cross-section of interest.

3. 3. The fiber cross-section inspection method according to claim 2, wherein an area range corresponding to a region including a length of 50% or less of a line segment extending from the fiber center toward the fiber outline is masked in the fiber cross-sections of the specimen image and the inspection image.

4. 3. The fiber cross-section inspection method according to claim 2, wherein an area range corresponding to a region including a length of 50% or less of a line segment extending from the outline of the fiber toward the center of the fiber in the fiber cross-section of the specimen image and the inspection image is masked.

Images