Fiber segmentation method
Through edge detection and binarization processing technology, the fiber profile is extracted and the fiber skeleton is separated, which solves the problem of fiber interlacing and difficult segmentation in textile composition detection, and realizes automatic fiber segmentation and automation of textile composition detection.
Patent Information
- Application Number
- CN202110390974.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-04-12
- Publication Date
- 2025-03-18
- Estimated Expiration
- 2041-04-12
AI Technical Summary
The prior art is difficult to automate textile component detection, mainly due to the crisscrossing of fibers, difficult to segment and identify, resulting in high artificial participation.
The fiber profile is extracted by edge detection method, and the fiber contour is determined through Gaussian adaptive thresholding and fixed threshold binarization. Combined with contour search and data cleaning, the fiber skeleton and intersection points are determined, and the cross fiber and free fiber are separated.
It realizes automatic segmentation and identification of fibers, opens up the key links in automatic textile component testing, and improves detection efficiency and accuracy.
Smart Images
Figure CN113192010B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of textile component detection, in particular to a fiber segmentation method. Background Art
[0002] At present, traditional textile component testing is carried out manually according to industry standards. Due to the high degree of manual involvement, automated component testing cannot be achieved. Reducing manual involvement in textile component testing is a key step in achieving automated testing. However, due to the filming process, the fibers in the pictures taken by the camera are crisscrossed and difficult to segment, and cannot be used for identification and component calculation. In summary, it is urgent to study methods that can automatically segment fibers to complete the key link of automated textile component testing. Summary of the invention
[0003] The object of the present invention is to provide a fiber splitting method comprising the following steps:
[0004] 1) Use edge detection method to extract the contours of all fibers in the image to be processed and obtain the fiber contour image.
[0005] The steps to obtain a fiber profile image include:
[0006] 1.1) Use the Gaussian adaptive thresholding method to binarize the image to be processed to obtain a preliminary binarized image.
[0007] 1.2) Perform data processing on the preliminary binary image, including filtering and noise reduction.
[0008] 1.3) The image obtained in step 2) is binarized using a fixed threshold binarization method to obtain an improved binarized image.
[0009] 1.4) Perform contour search and data cleaning on the improved binary image to obtain the fiber contour image.
[0010] The rule for data cleaning is: delete the data with an area smaller than the threshold s min silhouette.
[0011] 2) Extract the fiber skeleton in the fiber contour image and determine the center position of the fiber to obtain a fiber skeleton image.
[0012] The fiber connected area in the fiber outline image is thinned until it is thinned into a line with a pixel value. The line is the central axis skeleton of the fiber.
[0013] The steps of extracting the fiber skeleton include:
[0014] 2.1) Scan each point in turn from top to bottom and from left to right. The point currently scanned in the fiber connected area is recorded as (x, y). Determine the number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area. If the number of layers is 0, then the number of upper encirclement layers of point (x, y) in the fiber connected area is 1. If the minimum number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area is k, then the number of upper encirclement layers F(x, y) of point (x, y) in the fiber connected area is k+1. k≥1.
[0015] The number of upper enclosing layers F(x,y) of a point (x,y) in a fiber-connected region is expressed as follows:
[0016]
[0017] 2.2) Scan each point in turn from bottom to top and from right to left. The point currently scanned in the fiber connected area is recorded as (x, y). Determine the number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area. If the number of layers is 0, then the number of lower surrounding layers of point (x, y) in the fiber connected area is 1. If the minimum number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area is k, then the number of lower surrounding layers G(x, y) of point (x, y) in the fiber connected area is k+1. k≥1.
[0018] The number of lower enclosing layers G(x,y) of a point (x,y) in the fiber-connected region is expressed as follows:
[0019]
[0020] 2.3) Combine the upper and lower enclosing layers, and take the minimum of the upper and lower enclosing layers as the layer number L(x,y) of the current pixel (x,y), that is:
[0021] L(x,y)=Min(F(x,y),G(x,y)) (3)
[0022] 2.4) Keep the area with the highest number of layers surrounding the center of the connected region to obtain the skeleton of the connected region.
[0023] 3) According to the fiber skeleton image, determine the fiber intersection points and intersection position coordinates of different fibers.
[0024] The steps to determine fiber intersections include:
[0025] 3.1) The fiber skeleton image is convolved using the mean convolution kernel method to obtain a hotspot map of fiber intersections.
[0026] 3.2) Determine the fiber intersection points and intersection position coordinates based on the heat map of the fiber intersection points.
[0027] 4) Separate the crossing fibers in the fiber outline image.
[0028] 4.1) Find the intersection area in the fiber contour image based on the location of the intersection point.
[0029] 4.2) With the intersection point as the center, generate a circle with a radius of r and the color of the fiber outline image background color. This circle is the circle with the smallest area that can separate the crossed fibers.
[0030] 4.3) Erase the circle generated in step 4.2) in the fiber contour image to separate the crossed fibers.
[0031] 5) Searching for free fibers connected to the intersections and segmenting the free fibers, thereby completing the segmentation of the fibers in the processed image.
[0032] The step of segmenting the free fibers comprises:
[0033] 5.1) Search for free fibers connected to the intersection.
[0034] 5.2) Enlarge the intersection where the free fibers are connected and obtain a circular ring with the intersection as the center.
[0035] 5.3) Using the circular ring generated in step 5.2), all free fibers connected to the intersection are segmented to obtain fiber segments.
[0036] 5.4) Repeat steps 5.1) to 5.3) to traverse all intersections to obtain several fiber sections.
[0037] 5.5) Classify the fiber segments, and group the fiber segments belonging to the same fiber into one category. If the number of fiber segments in a category of fibers is greater than 1, only the longest fiber segment is retained.
[0038] The steps to classify fiber segments include:
[0039] 5.5.1) Obtain the outer edge contour of the fiber segment.
[0040] 5.5.2) Perform a straight line fitting on the outer edge contour of the fiber segment to obtain the straight line slope of the fiber segment.
[0041] 5.5.3) The straight line slope error is less than the preset threshold k max The fiber sections are grouped together.
[0042] 5.6) All retained fiber segment contours are merged to obtain free fiber contours.
[0043] 5.7) Generate a minimum circumscribed rectangle of the free fiber outline, and segment the free fibers according to the minimum circumscribed rectangle.
[0044] It is worth noting that the present invention first extracts the outlines of all fibers in the image through an edge detection method, and performs a binarization process to determine the approximate positions of all fibers. Then, the skeleton of the fiber is extracted based on the obtained binary image to determine the center position of the fiber in the image. Then, based on the overlap of the skeleton, the intersection position of different fibers is determined, and all the intersection points in the image are determined. Afterwards, based on the obtained intersection position, the point is gradually enlarged to become a circle of the background color until the intersecting fibers are completely separated into two or more different sections. Then, based on the intersection, the free fibers connected to it are searched, and it is determined whether they are the same free fibers based on the different directions of the fiber pointing, and the shorter section of the fiber is ignored. Finally, the fibers are completely segmented according to the contour extraction algorithm.
[0045] The technical effect of the present invention is unquestionable. The present invention can separate overlapping and crossing fibers in images taken with a microscope, thus opening up a key link for automated inspection of textiles. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] Figure 1 Schematic diagram of the fibers obtained by segmentation;
[0047] Figure 2 The picture to be processed;
[0048] Figure 3 Output image for adaptive thresholding;
[0049] Figure 4 To improve the adaptive thresholding of the output image;
[0050] Figure 5 Output image for fiber skeleton extraction;
[0051] Figure 6 It is the fiber heat map;
[0052] Figure 7 Schematic diagram of cross-fiber separation;
[0053] Figure 8 Schematic diagram of free fiber screening. DETAILED DESCRIPTION
[0054] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.
[0055] Embodiment 1:
[0056] See also Figures 1 to 8 , a fiber segmentation method, comprising the following steps:
[0057] 1) Use edge detection method to extract the contours of all fibers in the image to be processed and obtain the fiber contour image.
[0058] The steps to obtain a fiber profile image include:
[0059] 1.1) Use the Gaussian adaptive thresholding method to binarize the image to be processed to obtain a preliminary binarized image.
[0060] 1.2) Perform data processing on the preliminary binary image, including filtering and noise reduction.
[0061] 1.3) The image obtained in step 2) is binarized using a fixed threshold binarization method to obtain an improved binarized image.
[0062] 1.4) Perform contour search and data cleaning on the improved binary image to obtain the fiber contour image.
[0063] The rule for data cleaning is: delete the data with an area smaller than the threshold s min The contour of the min is the preset area threshold.
[0064] 2) Extract the fiber skeleton in the fiber contour image and determine the center position of the fiber to obtain a fiber skeleton image.
[0065] The fiber connected area in the fiber contour image is thinned until it is thinned into a line with a pixel value. The line is the central axis skeleton of the fiber. A fiber connected area is composed of several fibers that are in contact with each other in the image to be processed or free fibers that are not in contact with other fibers.
[0066] The steps of extracting the fiber skeleton include:
[0067] 2.1) Scan each point in turn from top to bottom and from left to right. The point currently scanned in the fiber connected area is recorded as (x, y). Determine the number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area. If the number of layers is 0, then the number of upper encirclement layers of point (x, y) in the fiber connected area is 1. If the minimum number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area is k, then the number of upper encirclement layers F(x, y) of point (x, y) in the fiber connected area is k+1. k≥1.
[0068] The number of upper enclosing layers F(x,y) of a point (x,y) in a fiber-connected region is expressed as follows:
[0069]
[0070] 2.2) Scan each point in turn from bottom to top and from right to left. The point currently scanned in the fiber connected area is recorded as (x, y). Determine the number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area. If the number of layers is 0, then the number of lower surrounding layers of point (x, y) in the fiber connected area is 1. If the minimum number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area is k, then the number of lower surrounding layers G(x, y) of point (x, y) in the fiber connected area is k+1. k≥1.
[0071] The number of lower enclosing layers G(x,y) of a point (x,y) in the fiber-connected region is expressed as follows:
[0072]
[0073] 2.3) Combine the upper and lower enclosing layers, and take the minimum of the upper and lower enclosing layers as the layer number L(x,y) of the current pixel (x,y), that is:
[0074] L(x,y)=Min(F(x,y),G(x,y)) (3)
[0075] 2.4) Keep the area with the highest number of layers surrounding the center of the connected region to obtain the skeleton of the connected region.
[0076] 3) According to the fiber skeleton image, determine the fiber intersection points and intersection position coordinates of different fibers.
[0077] The steps to determine fiber intersections include:
[0078] 3.1) The fiber skeleton image is convolved using the mean convolution kernel method to obtain a hotspot map of fiber intersections.
[0079] 3.2) Determine the fiber intersection points and intersection position coordinates based on the heat map of the fiber intersection points.
[0080] 4) Separate the crossing fibers in the fiber outline image.
[0081] 4.1) Find the intersection area in the fiber contour image based on the location of the intersection point.
[0082] 4.2) With the intersection point as the center, generate a circle with a radius of r and the color of the fiber outline image background color. This circle is the circle with the smallest area that can separate the crossed fibers.
[0083] The steps of generating a circle with a radius of r and a color equal to the background color of the fiber outline image include:
[0084] 4.2.1) With the intersection point as the center, generate a circle with a radius of r and the color of the fiber outline image background color; the initial value of r is r0; the value of r0 should be as small as possible.
[0085] 4.2.2) Determine whether the generated circle can separate the cross fibers. If not, update r = r + Δr and return to step 4.2.1). If yes, proceed to step 4.2.3); Δr is the radius increase. The value of Δr should be as small as possible.
[0086] 4.2.3) Update the radius r = r - Δr' and generate a circle with this radius. The value of Δr' is not greater than Δr / 2.
[0087] 4.2.4) Determine whether the generated circle can separate the crossed fibers. If not, end the process. If so, return to step 4.2.3).
[0088] 4.3) Erase the circle generated in step 4.2) in the fiber contour image to separate the crossed fibers.
[0089] 5) Searching for free fibers connected to the intersections and segmenting the free fibers, thereby completing the segmentation of the fibers in the processed image.
[0090] The step of segmenting the free fibers comprises:
[0091] 5.1) Search for free fibers connected to the intersection.
[0092] 5.2) Enlarge the intersection where the free fibers are connected and obtain a circular ring with the intersection as the center.
[0093] 5.3) Using the circular ring generated in step 5.2), all free fibers connected to the intersection are segmented to obtain fiber segments.
[0094] 5.4) Repeat steps 5.1) to 5.3) to traverse all intersections to obtain several fiber sections.
[0095] 5.5) Classify the fiber segments, and group the fiber segments belonging to the same fiber into one category. If the number of fiber segments in a category of fibers is greater than 1, only the longest fiber segment is retained.
[0096] The steps to classify fiber segments include:
[0097] 5.5.1) Obtain the outer edge contour of the fiber segment.
[0098] 5.5.2) Perform a straight line fitting on the outer edge contour of the fiber segment to obtain the straight line slope of the fiber segment.
[0099] 5.5.3) The straight line slope error is less than the preset threshold k max The fiber sections are grouped together.
[0100] 5.6) All retained fiber segment contours are merged to obtain free fiber contours.
[0101] 5.7) Generate a minimum circumscribed rectangle of the free fiber outline, and segment the free fibers according to the minimum circumscribed rectangle.
[0102] Embodiment 2:
[0103] The fiber segmentation method comprises the following steps:
[0104] 1) Extract fiber area
[0105] The present invention uses an improved adaptive thresholding operation to extract the fiber area. First, a Gaussian adaptive thresholding operation is used to perform binarization processing on the image to obtain a preliminary binary image. Then, a low-pass filter is used to filter the obtained preliminary binary image to remove high-frequency noise. Then, a fixed threshold binarization operation is performed to obtain a further improved binary image. Finally, the image is contour searched, and contours with an area less than a certain value are eliminated to obtain the final retained contours.
[0106] 2) Cross-fiber separation
[0107] 2.1) Determine the fiber skeleton
[0108] According to the obtained binary image of the fiber, the fiber area is thinned using the thinning technology to obtain the skeleton of the fiber. The thinning technology is to thin the connected area in the binary image until it is thinned into a line of one pixel value, which is the central axis skeleton of the fiber. The fast thinning algorithm adopted by the present invention repeatedly iterates and retrieves to obtain the final skeleton. The iterative steps are as follows:
[0109] 2.1.1) From top to bottom, from left to right
[0110] Scan each point from top to bottom and from left to right. Let the point in the current connected area be (x, y) and determine the number of layers of the four points to the left, upper left, upper and upper right in the connected area. If none of the above four points are in the connected area, the number of layers of (x, y) in the connected area is 1. If the minimum number of layers of the above four points in the connected area is k, the number of layers of (x, y) in the connected area is k+1. The formula is as follows:
[0111]
[0112] 2.1.2) From bottom to top, from right to left
[0113] Scan each point in turn from bottom to top and from right to left. Let the point in the current connected area be (x, y) and determine the number of layers of its right, lower right, lower and lower left in the connected area. If none of the above four points are in the connected area, the number of layers of (x, y) in the connected area is 1. If the minimum number of layers of the above four points in the connected area is k, the number of layers of (x, y) in the connected area is k+1. The formula is as follows:
[0114]
[0115] 2.1.3) Combine the upper and lower enclosing layers
[0116] Combine the upper enclosing layers with the lower enclosing layers, and take the lowest layer of the upper enclosing layers and the lower enclosing layers as the layer number of the current pixel. The formula is as follows:
[0117] L(x,y)=Min(F(x,y),G(x,y)) (3)
[0118] 2.1.4) Keep the area with the highest number of layers surrounding the center of the connected region to obtain the skeleton of the connected region.
[0119] 2.2) Find fiber intersections
[0120] According to the obtained fiber skeleton image, a 3x3 mean convolution kernel is used to convolve and obtain the hotspot map of the intersection. Finally, the intersection of the fibers is determined directly based on the hotspot map of the intersection.
[0121] 2.3) Separate the cross fibers according to the intersection point
[0122] By traversing all intersections, each intersection is enlarged and the crossed fibers are separated. According to the location of the intersection, the intersection area is found in the binary image, and then the intersection is completely erased by enlarging the circle with the same color as the background to achieve the separation of the crossed fibers.
[0123] 3) Separate the fibers and reassemble them
[0124] According to different intersections, search for all free fibers connected to the intersections in the image. Then further magnify the intersection to obtain a ring originating from the intersection. Then use the ring to segment all free fibers connected to the intersection to obtain a subsection of the fiber. Finally, analyze all the subsections to determine the free fibers that are the same fiber and combine them to obtain a complete fiber.
[0125] The outer edge contours of different fiber segments are obtained according to the outer edges, and then the positions of the two outer edge contours of the fiber are fitted with a straight line. Finally, the slopes of the straight lines of different fiber segments are judged based on the maximum bending degree of the fiber. If they are within a certain range, they are one fiber, otherwise they are not one fiber. For the same fiber segment, the shorter one is ignored and the longer one is retained.
[0126] 4) Split fibers
[0127] According to the results of fiber segment retention, all free fibers are judged. Finally, the contours of different free fibers are merged, and the minimum circumscribed rectangle of the obtained contours is calculated. And according to the minimum circumscribed rectangle, the fiber is completely segmented.
[0128] Embodiment 3:
[0129] The fiber segmentation method comprises the following steps:
[0130] 1) Get the original image to be processed, such as Figure 2 shown.
[0131] 1) Extract the fiber area, and obtain the adaptive threshold processing output image and the improved adaptive threshold processing output image, respectively. Figure 3 and Figure 4 shown.
[0132] 2) Cross-fiber separation
[0133] 2.1) Fiber skeleton extraction, such as Figure 5 shown.
[0134] 2.2) Find the intersection
[0135] According to the above skeleton, use a 3x3 convolution kernel with a convolution kernel value of 1. Convolve the skeleton image to get a heat map, such as Figure 6 Finally, according to the set threshold, find the position in the image that is greater than the threshold, and determine the position as the intersection of the fibers.
[0136] 2.3) Separate fibers based on crossover points
[0137] According to the determined fiber intersections, a circle is then drawn at the intersections to mask the fibers and separate them. If the fiber diameters vary greatly, an adaptive method can be used. In this embodiment, because the fiber diameters are relatively uniform, the radius of the circle is set to 10, and a mask image of free fibers is obtained, as shown in FIG. Figure 7 shown.
[0138] 2.4) Free fiber screening
[0139] According to the position of the intersection, search for the free fibers connected to it. Then, according to the feature selection method in the proposed method, obtain the characteristic straight lines of different free fibers. Finally, according to the slope of the straight line, determine whether it is one. In this embodiment, the allowed deviation of the slope is set to 3. Finally, the following mask image after screening is obtained, as shown in Figure 8 shown.
[0140] 4) Split fibers
[0141] The fibers are segmented according to the minimum circumscribed rectangle of the free fibers. For the convenience of display, in this embodiment, all the segmented fibers are marked with matrix frames in the original image, such as Figure 1 shown.
Claims
1. A fiber splitting method, characterized in that: The following steps are involved: 1) Using edge detection method to extract the contours of all fibers in the image to be processed, and obtain a fiber contour image; 2) Extract the fiber skeleton in the fiber contour image and determine the center position of the fiber to obtain a fiber skeleton image; 3) According to the fiber skeleton image, determine the fiber intersection points and intersection position coordinates of different fibers; 4) Separate the crossed fibers in the fiber outline image; 5) searching for free fibers connected to the intersections, and segmenting the free fibers, thereby completing the segmentation of the fibers in the processed image; The steps of extracting the fiber skeleton include: 2.1) Scan each point in turn from top to bottom and from left to right; the point currently scanned in the fiber connected area is recorded as (x, y), and the number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area is determined. If the number of layers is 0, the number of upper encirclement layers of point (x, y) in the fiber connected area is 1. If the minimum number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), (x, y-1) in the fiber connected area is k, then the number of upper encirclement layers F(x, y) of point (x, y) in the fiber connected area is k+1; k≥1; The number of upper enclosing layers F(x,y) of a point (x,y) in a fiber-connected region is expressed as follows: 2.2) Scan each point in turn from bottom to top and from right to left; the point currently scanned in the fiber connected area is recorded as (x, y), and the number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), and (x, y-1) in the fiber connected area is determined. If the number of layers is 0, the number of lower surrounding layers of point (x, y) in the fiber connected area is 1. If the minimum number of layers of points (x-1, y+1), (x-1, y), (x-1, y-1), and (x, y-1) in the fiber connected area is k, then the number of lower surrounding layers G(x, y) of point (x, y) in the fiber connected area is k+1; k≥1; The number of lower enclosing layers G(x,y) of a point (x,y) in the fiber-connected region is expressed as follows: 2.3) Combine the upper and lower enclosing layers, and take the minimum of the upper and lower enclosing layers as the layer number L(x,y) of the current pixel (x,y), that is: L(x,y)=Min(F(x,y),G(x,y)) (3) 2.4) Keep the area with the highest number of layers surrounding the center of the connected area to obtain the skeleton of the connected area; The step of segmenting the free fibers comprises: 5.1) Search for free fibers connected to the intersection; 5.2) Enlarge the intersection where the free fibers are connected to obtain a circular ring centered at the intersection; 5.3) using the ring generated in step 5.2) to segment all free fibers connected to the intersection to obtain fiber segments; 5.4) Repeat steps 5.1) to 5.3) to traverse all intersections to obtain several fiber sections; 5.5) Classify the fiber segments, and classify the fiber segments belonging to the same fiber into one category; if the number of fiber segments in a category of fibers is greater than 1, only the longest fiber segment is retained; 5.6) Merge all the retained fiber segment contours to obtain the free fiber contour; 5.7) Generate a minimum circumscribed rectangle of the free fiber outline, and segment the free fibers according to the minimum circumscribed rectangle.
2. The fiber splitting method according to claim 1, characterized in that: The steps to obtain a fiber profile image include: 1) Use the Gaussian adaptive thresholding method to binarize the image to be processed to obtain a preliminary binarized image; 2) Perform data processing on the preliminary binary images, including filtering and noise reduction; 3) Using a fixed threshold binarization method, the image obtained in step 2) is binarized to obtain an improved binarized image; 4) Perform contour search and data cleaning on the improved binary image to obtain a fiber contour image.
3. The fiber splitting method according to claim 2, characterized in that: The rule for data cleaning is: delete the data with an area smaller than the threshold s min silhouette.
4. The fiber splitting method according to claim 1, characterized in that: The method for extracting the fiber skeleton is: thinning the fiber connected area in the fiber contour image until it is thinned into a line with a pixel value; the line is the central axis skeleton of the fiber.
5. The fiber splitting method according to claim 1, characterized in that: The steps to determine fiber intersections include: 1) The fiber skeleton image is convolved using the mean convolution kernel method to obtain a hotspot map of fiber intersections; 2) Determine the fiber intersection points and intersection position coordinates based on the heat map of the fiber intersection points.
6. The fiber splitting method according to claim 1, characterized in that: The steps for separating cross-fibers include: 1) Find the intersection area in the fiber contour image based on the location of the intersection point; 2) With the intersection point as the center, generate a circle with a radius of r and the color of the fiber outline image background color. This circle is the circle with the smallest area that can separate the crossed fibers; 3) Erase the circle generated in step 2) in the fiber contour image to separate the crossed fibers.
7. The fiber splitting method according to claim 1, characterized in that: The steps to classify fiber segments include: 1) Obtain the outer edge contour of the fiber segment; 2) Perform a straight line fitting on the outer edge contour of the fiber segment to obtain the straight line slope of the fiber segment; 3) The straight line slope error is less than the preset threshold k max The fiber sections are grouped together.
Citation Information
Patent Citations
Overlapped fiber image segmentation method and device, storage medium, and computer equipment
CN107256406A
Fibrous septum characteristic quantification method and apparatus
CN108230387A
Cited By
An electron microscope image processing method for fiber identification and volume fraction calculation
CN122637395A