Light stripe recovery and clustering matching method based on limited area growth

Abstract

The invention relates to a light stripe recovery and clustering matching method based on limited area growth, and belongs to the technical field of computer vision measurement. Comprising the following steps: step 1, obtaining stripe projection images after segmentation of a region of interest of a measured object under two camera visual angles; 2, obtaining a binary fringe image; 3, extracting a stripe center line of the binary stripe image through connected domain analysis to obtain a seed point coordinate of each stripe; step 4, obtaining stripe point coordinate data; 5, according to the stripe point coordinate data recovered in the step 4, calculating the center-of-mass coordinates of the stripes, and numbering the stripes in the left view and the right view; and step 6, comparing the fringe numbers in the left and right views, if the fringe numbers are not matched, increasing the area growth radius, and executing the step 4 and the step 5 again until the fringe numbers in the left and right views are matched. According to the invention, the problems of noise, fracture and deformation in the light stripe image can be effectively solved, and the problem of mismatching of multi-line laser stripes is avoided.

Description

Technical Field

[0001] The present invention relates to a light stripe restoration and cluster matching method based on restricted area growth, and belongs to the technical field of computer vision measurement. Background Art

[0002] In a multi-line laser-based 3D measurement system, multiple laser lines are projected onto the surface of the object being measured. A camera captures the projected fringe image, and then, through steps such as centerline extraction, epipolar constraints, and stereo matching, the 3D coordinates of the object's surface are calculated. However, factors such as the complex texture of the object's surface, highly variable morphology, interference from external light sources, and image noise often severely impact the fringe extraction process, resulting in lost, interlaced, broken, or noise-inducing fringes. These issues make fringe matching extremely difficult, leading to mismatched fringes in the left and right images, ultimately compromising the accuracy of the object's 3D reconstruction.

[0003] To address these issues, existing technologies typically use traditional image processing and clustering methods. However, these methods often perform poorly in stripe recovery and matching under complex backgrounds and lighting conditions. In particular, when stripes are broken or incomplete, the accuracy of accurately recovering and matching stripes is low. Therefore, overcoming the difficulties in stripe extraction and matching through more efficient image recovery and matching algorithms remains an important research issue in this field. Summary of the Invention

[0004] To address the difficulty of accurately clustering and matching fringes in existing binocular-based multi-line laser 3D measurement systems when fringes are broken, blurred, or wide, this paper proposes a light fringing recovery and cluster matching method based on restricted region growing. This method can recover and reconstruct damaged and missing multi-line fringes without requiring optical plane calibration, and also cluster and match fringes between left and right views. This method leverages the structural continuity of fringes and the physical consistency between left and right images, combining region growing and clustering. Even in images with complex noise and discontinuous fringes, this method effectively recovers and matches corresponding fringes in the left and right images, achieving stable fringing extraction and matching, and providing high-quality input data for subsequent 3D reconstruction.

[0005] The technical solution adopted by the present invention is: a light streak restoration and cluster matching method based on restricted area growth, comprising the following steps:

[0006] Step 1: Build a multi-line laser measurement model based on binocular vision to obtain fringe projection images of the region of interest of the object under test after segmentation from the perspectives of two cameras;

[0007] Step 2: The fringe projection image obtained in step 1 is blurred, broken, and widened to simulate the actual fringe deformation, and the processed fringe projection image is filtered and binarized to obtain a clear binary fringe image;

[0008] Step 3: Extract the stripe center lines of the binary stripe image in step 2 through connected domain analysis, and obtain the seed point coordinates of each stripe as the starting point for subsequent recovery;

[0009] Step 4: Based on the coordinates of each seed point, the restricted region growing algorithm is used to restore the stripes and obtain the stripe point coordinate data;

[0010] Step 5: Based on the restored fringe point coordinate data obtained in step 4, calculate the centroid coordinates of the fringe and number the fringe in the left and right views; the fringe with the same number in the left and right views is the corresponding matching fringe;

[0011] Step 6: Compare the number of stripes in the left and right views. If they do not match, increase the radius of the region growing and repeat steps 4 and 5 until the number of stripes in the left and right views matches.

[0012] Step 1 specifically includes the following steps:

[0013] When constructing a divergent multi-line laser measurement model based on binocular vision, the multi-line laser needs to be projected orthographically and the projected stripes need to be vertical or tilted in the view imaging;

[0014] When collecting pictures, a solid-color background paper with a color different from that of the object to be measured is placed on the measuring platform. The two cameras first collect a picture of the object to be measured without laser fringe projection without moving the object to be measured. Then, a picture of the object to be measured with structured light fringe projection is collected. By detecting and extracting edge features of the picture without laser fringe projection and performing image dilation on the extracted edge contour, an accurate closed image contour is obtained. The closed contour is then filled to obtain a mask image of the ROI area of the object to be measured, which is then superimposed on the fringe projection image of the object to be measured to obtain a fringe projection image after segmentation of the region of interest.

[0015] Specifically, the binarization process in step 2 refers to processing the stripes in the image into clear white and the background into black, and using the thresholding method of OpenCV to convert the widened, blurred, broken and filtered image into a binary image.

[0016] The specific steps of step 3 are as follows:

[0017] 1) Connected domain analysis: The binary stripe image obtained in step 2 is processed using the connected domain analysis method to identify the independent region of each stripe. The pixel coordinates of each stripe are obtained using the cv2.connectedComponents() method in OpenCV.

[0018] 2) Extraction of the center line of the stripe: Analyze each stripe and calculate the center line position of the stripe. The specific method is to calculate the mean value of the pixels in each stripe area and obtain the centroid or center point coordinates (x s ,y s );

[0019] 3) Preservation of seed point coordinates: Save the centroid or center point coordinates (x s ,y s ) are saved as seed points, and the coordinates of these seed points will be used as the starting points of the subsequent restricted region growing algorithm.

[0020] The restricted region growing algorithm in step 4 specifically includes:

[0021] (1) Perform gradient calculation on the input image I(x,y) and use the Sobel operator in the horizontal and vertical directions to calculate the gradient amplitude of the image:

[0022]

[0023] Then the gradient amplitude G(x,y) can be expressed as:

[0024]

[0025] Among them, G x , G y are the gradients of a certain stripe image in the x and y directions respectively, G(x,y) is the total gradient amplitude of the stripe on the image, and the gradient information is used to control the growth of the stripe;

[0026] (2) Initialize region growing parameters:

[0027] Each stripe seed point S obtained in step 3 is S=(x s ,y s ) as the center, set the maximum and minimum values of the growth window to W min and W max , the real-time window W size is dynamically adjusted to the range:

[0028] W min ≤W≤W max ,W∈{2k+1|k∈N}

[0029] Where W min : The minimum size allowed for the growth window;

[0030] W max : The maximum size allowed for the growth window;

[0031] W: The size of the current growing window, which must be an odd number 2k+1, where k is a natural number;

[0032] (3) Region growing from the seed point:

[0033] The current pixel to be expanded is defined as P c =(x c ,y c ), with P c A growth window of size W×W is established as the center, and the gradient amplitude area within the growth window is defined as:

[0034]

[0035] R W (P c ): Point P c The set of all gradient amplitude points within the W×W window centered at ;

[0036] G(i,j): the gradient magnitude of the image at coordinate (i,j);

[0037] x c ,y c : The horizontal and vertical coordinates of the center point of the current window;

[0038] Half the length of the window, rounded down;

[0039] And calculate the variance of the gradient magnitude within the window:

[0040]

[0041] in Window R W (P c ) Variance of the internal gradient amplitude

[0042] μ G : Window R W (P c )The average value of the gradient amplitude

[0043] W 2 : The total number of pixels contained in the window;

[0044] like When , it indicates that the gradient changes gently within the region and the window can be appropriately expanded:

[0045] W=min(W+2,W max )

[0046] like When the gradient changes drastically in the region, the window should be reduced:

[0047] W=max(W-2,W min );

[0048] T g : Gradient threshold, used to determine whether the pixel has sufficient gradient features;

[0049] (4) Pixel inclusion conditions:

[0050] For the pixel point P in the window n =(x n ,y n ), if:

[0051] Pixel I(x n ,y n ) has a non-zero grayscale value and is not processed;

[0052] Gradient amplitude G(x n ,y n ) is greater than a preset threshold;

[0053] After adding the new point, calculate the three local continuous points P of the stripe i-2 ,P i-1 ,P i The curvature C formed is calculated using the curvature calculation formula:

[0054]

[0055] If C≤T c , then P n Add to the region growing queue and mark it as processed; otherwise, abandon the point to avoid abnormal distortion of the stripe boundary. c : Curvature threshold, used to control the smoothness of the stripe shape;

[0056] (5) Iterative loop of region growing:

[0057] Continuously update the window center point P c , perform the above iterative process of (3) and (4) for each new point in the queue until the queue is empty, that is, the restoration of the stripe area is completed.

[0058] The present invention provides the following beneficial effects: The proposed restricted region growing restoration method can address issues such as breakage, blurring, intersection, and widening in real-world laser stripe images. Through precise seed-driven restoration and clustering, it avoids the traditional methods' strong reliance on stripe continuity and interference from noise points. Furthermore, the method is independent of light plane calibration, exhibits strong adaptability, and can achieve stable stripe restoration and matching on irregular surfaces or in complex measurement scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 is a system flow chart of the present invention;

[0060] Figure 2 Schematic diagram of the measurement system of the present invention;

[0061] Figure 3 It is a schematic diagram of region segmentation of the present invention;

[0062] Figure 4 Establishing a schematic diagram for the confined region growth of the present invention;

[0063] Figure 5 This is a projection of blurred, broken, and enlarged light stripes

[0064] Figure 6 Schematic diagram of the experimental results of the present invention. DETAILED DESCRIPTION

[0065] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0066] Example 1: The present invention provides a method for recovering and clustering light stripes using restricted region growth. First, the object to be measured is segmented from the carrier plane. Second, the extracted centerline stripes are clustered using restricted region growth. Finally, based on the sequential consistency constraint, the stripe centroid coordinates are used to number the clustered stripes in the left and right views. Stereo matching is performed for stripes with the same number. The structural block diagram is shown below. Figure 1 shown.

[0067] like Figure 2 Figure 2 shows a schematic diagram of a measurement system for clustering and matching multi-line laser stripes based on custom window iteration, including two industrial cameras 1, a divergent multi-line laser 2, an object to be measured 3, a background plate 4, a bracket 5, and a workbench 6. The two cameras have identical parameters, are approximately parallel in the horizontal direction, and are symmetrically arranged on bracket 5 relative to the multi-line laser.

[0068] The method for clustering and matching multi-line laser stripes based on custom window iteration includes the following steps:

[0069] Step 1: Construct a multi-line laser measurement model based on binocular vision to obtain the fringe projection image of the region of interest of the object under test after segmentation under the perspectives of two cameras.

[0070] The multi-line laser is projected orthogonally and the projected stripes need to be vertical or tilted in the view imaging. When collecting pictures, in order to accurately segment the area, it is necessary to place a solid color background paper on the measuring platform that is different in color from the object to be measured. The present invention provides a background material used on the measuring platform to effectively reduce the influence of background noise. The material can be clearly distinguished from the object to be measured, and can be a material of various colors and textures. The two cameras first capture a picture of the object to be measured without laser stripe projection, without moving the object to be measured, and then capture a picture of the object to be measured with structured light stripe projection. By detecting and extracting the edge features of the picture without laser stripe projection, and performing image expansion on the extracted edge contour, an accurate closed image contour is obtained, and then the closed contour is filled to obtain a mask map of the ROI area of the object to be measured. It is superimposed with the image of the object to be measured by stripe projection to obtain a stripe projection image after the region of interest is segmented, such as Figure 3 shown. Figure 3 The figure includes 4 sub-images. The upper left and upper right images are full projection images of the object under test and the background with structured light stripes projected on a black background, captured by the binocular camera; the lower left and lower right images are object projection images after the upper left and upper right images are masked through the ROI area of the object under test (the light stripes in the black background area are removed).

[0071] Step 2: The fringe projection image obtained in step 1 is blurred, broken, and widened to simulate the actual fringe deformation, and the processed fringe projection image is filtered and binarized to obtain a clear binary fringe image.

[0072] Furthermore, binarization involves converting the image's stripes to a clear white background and a black background. This is done by using OpenCV's thresholding method to convert the widened, blurred, broken, and filtered image into a binary image. Specifically, the coordinates of each pixel are plotted with a value of 255 on an image with the same format as the original image and all pixel values set to 0, resulting in a binary image of the centerline stripes.

[0073] Step 3: Perform connected region extraction analysis on the centerline binary image generated in step 2, extract all connected regions in the image (i.e. complete stripes or broken fragments), and calculate the centroid of the pixel coordinate set in each connected region. The centroid position is the seed point coordinate of the stripe. These points will serve as the starting points for subsequent region growth, such as Figure 4The principle diagram of "restoration of light stripes grown in a restricted area" is shown in the figure. The algorithm starts with the seed point P0 on the broken stripe, and gradually grows the stripe structure forward by analyzing the image gradient and curvature information in the local window. At each step, if the gradient at the current point changes slowly, the window (such as W2, W3) is appropriately expanded to obtain more contextual information; if the gradient fluctuates violently or the curvature is too large (such as C>Tc), the window is reduced to avoid the stripes from being incorrectly extended or deviating from the original path. Finally, under the control of dynamically adjusting the window size, the algorithm achieves the complete restoration of stripes in distorted areas such as broken and blurred areas, and outputs a continuous and smooth "complete stripe after restoration" trajectory. Specifically:

[0074] 1) Connected domain analysis: The binary stripe image obtained in step 2 is processed using the connected domain analysis method to identify the independent region of each stripe. The pixel coordinates of each stripe are obtained using the cv2.connectedComponents() method in OpenCV.

[0075] 2) Extraction of the center line of the stripe: Analyze each stripe and calculate the center line position of the stripe. The specific method is to calculate the mean value of the pixels in each stripe area and obtain the centroid or center point coordinates (x s ,y s );

[0076] 3) Preservation of seed point coordinates: Save the centroid or center point coordinates (x s ,y s ) are saved as seed points, and the coordinates of these seed points will be used as the starting points of the subsequent restricted region growing algorithm.

[0077] Step 4: Based on the coordinates of each seed point, the restricted region growing algorithm is used to restore the stripes and obtain the stripe point coordinate data. The specific steps are: using the local gradient information of the image, dynamically adjusting the size and shape of the growing window to ensure the restoration of the stripe shape and avoid interlacing and breaking between stripes.

[0078] The restricted region growing algorithm in step 4 specifically includes:

[0079] (1) Perform gradient calculation on the input image I(x,y) and use the Sobel operator in the horizontal and vertical directions to calculate the gradient amplitude of the image:

[0080]

[0081] Then the gradient amplitude G(x,y) can be expressed as:

[0082]

[0083] Among them, G x , Gy are the gradients of the image in the x and y directions, respectively, and the gradient information is used to control the growth of the stripes.

[0084] (2) Initialize region growing parameters:

[0085] Each stripe seed point S obtained in step 3 is S=(x s ,y s ) as the center, set the initial growth window size to W min and W max Define the gradient threshold as T g , the curvature threshold is T c , the real-time window W size is dynamically adjusted to the range:

[0086] W min ≤W≤W max ,W∈{2k+1|k∈N}

[0087] (3) Region growing from the seed point:

[0088] The current pixel to be expanded is defined as P c =(x c ,y c ), with P c A growth window of size W×W is established as the center, and the gradient amplitude area within the growth window is defined as:

[0089]

[0090] And calculate the variance of the gradient magnitude within the window:

[0091]

[0092] like When , it indicates that the gradient changes gently within the region and the window can be appropriately expanded:

[0093] W=min(W+2,W max )

[0094] like When the gradient changes drastically in the region, the window should be reduced:

[0095] W=max(W-2,W min )

[0096] (4) Pixel inclusion conditions:

[0097] For the pixel point P in the window n =(x n ,y n ), if:

[0098] Pixel I(xn ,y n ) has a non-zero grayscale value and is not processed;

[0099] Gradient amplitude G(x n ,y n ) is greater than a preset threshold;

[0100] After adding the new point, calculate the three local continuous points P of the stripe i-2 ,P i-1 ,P i The curvature C formed is calculated using the curvature calculation formula:

[0101]

[0102] If C≤T c , then P n Add the point to the region growing queue and mark it as processed; otherwise, abandon the point to avoid abnormal distortion of the stripe boundary.

[0103] (5) Iterative loop of region growing:

[0104] Continuously update the window center point P c , perform the above iterative process of (3) and (4) for each new point in the queue until the queue is empty, that is, the restoration of the stripe area is completed.

[0105] like Figure 5 As shown, Figure 5 The paper demonstrates common types of laser stripe image distortion in real-world measurement scenarios, including broken, blurred, wide, missing, curved, and fuzzy stripes, as well as sharp deflections caused by surface deformation. These distortions reflect the multiple interference factors encountered during the projection of multi-line lasers onto complex surfaces. These distortions can severely interfere with the accurate extraction of the stripe centerline and impact the accuracy of subsequent 3D reconstruction. Figure 6 Then Figure 5 The image shows the restored image after processing with the constrained region growing algorithm proposed in this paper. As can be seen in the figure, the previously broken, blurred, and missing stripes have been effectively restored. The stripe centerlines have been successfully extracted and marked with a solid red line. The green dashed box represents the adaptively adjusted growth window, and the green dots are the initial seed points for each stripe. This figure demonstrates the method's high robustness and ability to accurately restore stripe structure under complex distortion conditions. It significantly improves the integrity and continuity of the stripes, providing an accurate input basis for subsequent cluster numbering and dual-view matching.

[0106] Step 5: Based on the restored stripe point coordinate data obtained in step 4, calculate the centroid coordinates of the stripes and number the stripes in the left and right views; stripes with the same number in the left and right views are the corresponding matching stripes.

[0107] The centroid coordinates of each stripe (ave x ,ave y ) and the calculation formula of the Euclidean distance dis from the centroid coordinate to the pixel coordinate origin is as follows:

[0108]

[0109] Among them, ave x 、ave y are the horizontal and vertical coordinates of the stripe centroid, and g is the number of stripes in the container v after clustering. j The number of point coordinates contained in x l 、y l Represents the horizontal and vertical coordinate values of a point in the stripe, and dis is the Euclidean distance from the centroid coordinate to the pixel coordinate origin.

[0110] Step 6: Compare the number of stripes in the left and right views. If they do not match, increase the radius of the region growing and repeat steps 4 and 5 until the number of stripes in the left and right views matches.

[0111] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, such as the model parameters of the divergent multi-line laser, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be encompassed within the present invention. Any reference signs in the claims should not be construed as limiting the claim to which they relate.

[0112] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A light streak restoration and cluster matching method based on restricted region growth, characterized by: The following steps are involved: Step 1: Build a multi-line laser measurement model based on binocular vision to obtain fringe projection images of the region of interest of the object under test after segmentation from the perspectives of two cameras; Step 2: The fringe projection image obtained in step 1 is blurred, broken, and widened to simulate the actual fringe deformation, and the processed fringe projection image is filtered and binarized to obtain a clear binary fringe image; Step 3: Extract the stripe center lines of the binary stripe image in step 2 through connected domain analysis, and obtain the seed point coordinates of each stripe as the starting point for subsequent recovery; Step 4: Based on the coordinates of each seed point, the restricted region growing algorithm is used to restore the stripes and obtain the stripe point coordinate data; Step 5: Based on the restored fringe point coordinate data obtained in step 4, calculate the centroid coordinates of the fringe and number the fringe in the left and right views; the fringe with the same number in the left and right views is the corresponding matching fringe; Step 6: Compare the number of stripes in the left and right views. If they do not match, increase the radius of the region growing and repeat steps 4 and 5 until the number of stripes in the left and right views matches.

2. The light streak restoration and cluster matching method based on restricted area growth according to claim 1, characterized in that: Step 1 specifically includes the following steps: When constructing a divergent multi-line laser measurement model based on binocular vision, the multi-line laser needs to be projected orthographically and the projected stripes need to be vertical or tilted in the view imaging; When collecting pictures, a solid-color background paper with a color different from that of the object to be measured is placed on the measuring platform. The two cameras first collect a picture of the object to be measured without laser fringe projection without moving the object to be measured. Then, a picture of the object to be measured with structured light fringe projection is collected. By detecting and extracting edge features of the picture without laser fringe projection and performing image dilation on the extracted edge contour, an accurate closed image contour is obtained. The closed contour is then filled to obtain a mask image of the ROI area of the object to be measured, which is then superimposed on the fringe projection image of the object to be measured to obtain a fringe projection image after segmentation of the region of interest.

3. The light streak restoration and cluster matching method based on restricted area growth according to claim 1, characterized in that: The binarization process in step 2 refers to processing the stripes in the image into clear white and the background into black. The thresholding method of OpenCV is used to convert the widened, blurred, broken and filtered image into a binary image.

4. The light streak restoration and cluster matching method based on restricted area growth according to claim 1, characterized in that: The specific steps of step 3 are as follows: 1) Connected domain analysis: The binary stripe image obtained in step 2 is processed using the connected domain analysis method to identify the independent region of each stripe. The pixel coordinates of each stripe are obtained using the OpenCV method. 2) Extraction of the center line of the stripe: Analyze each stripe and calculate the center line position of the stripe. The specific method is to calculate the mean value of the pixels in each stripe area and obtain the centroid or center point coordinates (x s ,y s ); 3) Preservation of seed point coordinates: Save the centroid or center point coordinates (x s ,y s ) are saved as seed points, and the coordinates of these seed points will be used as the starting points of the subsequent restricted region growing algorithm.

5. The light streak restoration and cluster matching method based on restricted area growth according to claim 4, characterized in that: The method in OpenCV refers to the cv2.connectedComponents() method.

6. The light streak restoration and cluster matching method based on restricted area growth according to claim 1, characterized in that: The restricted region growing algorithm in step 4 specifically includes: (1) Perform gradient calculation on the input image I(x,y) and use the Sobel operator in the horizontal and vertical directions to calculate the gradient amplitude of the image: Then the gradient amplitude G(x,y) can be expressed as: Among them, G x , G y are the gradients of a certain stripe image in the x and y directions respectively, G(x,y) is the total gradient amplitude of the stripe on the image, and the gradient information is used to control the growth of the stripe; (2) Initialize region growing parameters: Each stripe seed point S obtained in step 3 is S=(x s ,y s ) as the center, set the maximum and minimum values of the growth window to W min and W max , the real-time window W size is dynamically adjusted to the range: IN min ≤W≤W max ,W∈{2k+1∣k∈N} Where W min : The minimum size allowed for the growth window; W max : The maximum size allowed for the growth window; W: The size of the current growing window, which must be an odd number 2k+1, where k is a natural number; (3) Region growing from the seed point: The current pixel to be expanded is defined as P c =(x c ,y c ), with P c A growth window of size W×W is established as the center, and the gradient amplitude area within the growth window is defined as: R W (P c ): Point P c The set of all gradient amplitude points within the W×W window centered at ; G(i,j): the gradient magnitude of the image at coordinate (i,j); x c ,y c : The horizontal and vertical coordinates of the center point of the current window; Half the length of the window, rounded down; And calculate the variance of the gradient magnitude within the window: in Window R W (P c ) Variance of the internal gradient amplitude μ G : Window R W (P c )The average value of the gradient amplitude W 2 : The total number of pixels contained in the window; like When , it indicates that the gradient changes gently within the region and the window can be appropriately expanded: W=min(W+2,W max ) like When the gradient changes drastically in the region, the window should be reduced: W=max(W-2,W min ); T g : Gradient threshold, used to determine whether the pixel has sufficient gradient features; (4) Pixel inclusion conditions: For the pixel point P in the window n =(x n ,y n ), if: Pixel I(x n ,y n ) has a non-zero grayscale value and is not processed; Gradient amplitude G(x n ,y n ) is greater than a preset threshold; After adding the new point, calculate the three local continuous points P of the stripe i-2 ,P i-1 ,P i The curvature C formed is calculated using the curvature calculation formula: If C≤T c , then P n Add to the region growing queue and mark it as processed; otherwise, abandon the point to avoid abnormal distortion of the stripe boundary. c : Curvature threshold, used to control the smoothness of the stripe shape; (5) Iterative loop of region growing: Continuously update the window center point P c , perform the above iterative process of (3) and (4) for each new point in the queue until the queue is empty, that is, the restoration of the stripe area is completed.

Images