Intelligent real-time extraction and three-dimensional reconstruction method for contour of ring forging formed by ring rolling

Abstract

The invention belongs to the technical field of photoelectric measurement, and particularly relates to an intelligent real-time extraction and three-dimensional reconstruction method for the contour of a ring forging formed through ring rolling. According to the method, through a model cascade thought, a prediction depth map is input into a semantic segmentation model, then an edge detection algorithm, morphological operation and the like are carried out on a segmentation result, online contour extraction of the ring part in the intelligent ring forging process is realized by applying a contour completion algorithm, and then reconstruction of a three-dimensional geometric model is realized. The method is suitable for annular forgings with various bottom surface shapes, and has good robustness. In addition, the method can effectively complement partial contour missing caused by equipment shielding, and the integrity and accuracy of contour data are ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of optoelectronic measurement, and particularly relates to a method for real-time extraction and three-dimensional reconstruction of the contour of an intelligent ring forging formed by rotary forging. Background Art

[0002] Ring forging (abbreviated as "ring forging") is a key metal processing technology, mainly used for manufacturing components with a ring-shaped cross-section. In this process, the metal blank is heated to a high temperature, and then plastic deformation is applied to it by means of a press or forging equipment, gradually expanding the diameter of the ring, reducing the cross-sectional thickness, while the height remains basically unchanged, finally forming the required ring-shaped part. During the ring forging process, accurately and real-time obtaining the contour information of the ring-shaped part is crucial for precisely controlling the deformation of the ring-shaped part, reducing shape deviation and dimensional error.

[0003] Currently, the surface contour measurement of high-temperature forgings (including ring-shaped parts) mostly uses laser scanning or multi-view stereo vision technology. Compared with the traditional close-range coordinate measuring technology, these non-contact measuring technologies not only improve the measurement accuracy but also effectively avoid the influence of high temperature and harsh environment on the equipment and workpieces. However, most of these methods are used for the measurement of static high-temperature forgings or forgings with simple shapes (such as circular). For example, laser scanning requires the forging to remain stationary or multiple scans are needed to ensure the integrity of the data, resulting in a long measurement time and making it difficult to achieve real-time measurement during the deformation process of the forging. The multi-view stereo vision technology, on the other hand, faces the problems of long algorithm processing time and poor robustness. Especially when the target object is partially occluded, it is difficult to accurately complete the contour filling. These limitations make real-time measurement and application under complex working conditions difficult.

[0004] Bokhabrine et al. developed a non-contact measurement system that uses two laser scanners to collect point clouds based on the time-of-flight principle, optimizes the scanner positions through a genetic algorithm, and uses three-dimensional segmentation and ICP algorithms to reconstruct the point clouds. At the same time, complete 3D data is collected by rotating the part. The measurement error of this system on hot metal parts is less than 8 mm, with relatively high accuracy, but the algorithm is limited to circular objects and has poor robustness.

[0005] The team led by Zhang Yucun has achieved many results in the field of high-precision measurement of ring forgings. In 2014, the team used laser triangulation combined with the PSO algorithm to calibrate the system and improve the diameter measurement accuracy of high-temperature cylindrical forgings, achieving a three-dimensional measurement error of less than 1 mm. In 2017, the team proposed a scanning method applicable to rotating forgings, with an outer diameter measurement error of less than 6 mm. In 2018, the team further developed a measurement system for complex ring forgings, using the Kalman filter method to compensate for errors and reconstructing the model through topological theory, achieving high-precision measurement with a flatness deviation of less than 1.74 mm. However, the team's research focused on forgings with cylindrical or circular interfaces, mainly concerned with the diameter or height of the forgings, and had not solved the problem of contour measurement of forgings with other shapes during the deformation process.

[0006] Zhou Yijun et al. established a binocular stereo vision measurement system, innovatively using feature lines to replace traditional feature points for 3D measurement of high-temperature forgings. By back-projection, the contour line of the object surface was obtained, avoiding the influence of feature point noise, simplifying image processing, and reducing costs. However, the measurement error of this system in the forging workshop was 0.79%, and the measurement time was 1.9128 seconds, still not meeting the time requirement for real-time measurement. Summary of the Invention

[0007] In view of this, to solve the problems existing in the prior art, the present invention provides a method for real-time extraction of the contour and three-dimensional geometric reconstruction of a ring forging during the intelligent ring rolling forming process.

[0008] To achieve the above object, the object of the present invention is to provide a method for real-time extraction of the contour and three-dimensional reconstruction of an intelligent ring rolling formed ring forging.

[0009] A method for real-time extraction of the contour and three-dimensional reconstruction of an intelligent ring rolling formed ring forging, which uses an industrial camera as an image data acquisition device, and through camera pose optimization and data communication, real-time perceives and obtains the geometric morphology information RGB image of the ring part during the intelligent ring forging process. Then, using the RGB image as input data, through a depth prediction model, a depth image is obtained in real-time; the depth image and semantic cue points are input into a semantic segmentation model to obtain the mask of the ring part in real-time; based on the mask result, morphological operations and contour recognition are performed, and through a clustering algorithm, the contour points of the inner and outer rings of the upper bottom surface of the ring part are obtained; and taking the front view of the ring part as input, the height of the ring part is obtained; finally, through algorithm optimization, the geometric three-dimensional reconstruction of the ring part is completed according to the contour points of the upper bottom surface of the ring part and the height of the ring part.

[0010] Considering that the existing techniques for high-temperature forging profile extraction and measurement are mainly applicable to forgings in a static state and cannot meet the real-time requirements of deformed forgings during the forging process. In addition, the existing real-time measurement algorithms have poor robustness, are usually only applicable to forgings with specific shapes, and fail to effectively solve the problem of profile completion for the parts of the forging blocked by the equipment during the processing. These limitations affect the application of the existing technologies in complex working conditions. Through the idea of model cascading, the present invention uses the predicted depth map as the input data of the semantic segmentation model. By performing edge detection algorithms and morphological operations on the segmentation results and applying the profile completion algorithm, it can realize the online profile extraction of the ring-shaped parts during the intelligent ring forging process, and then realize the reconstruction of the three-dimensional geometric model. It is applicable to ring forgings with various bottom shapes and has good robustness. In addition, it can effectively complete the missing part of the profile caused by equipment occlusion, ensuring the integrity and accuracy of the profile data.

[0011] Further, when the height of the ring-shaped part remains unchanged, the height of the ring-shaped part is obtained according to the initial parameters.

[0012] Further, for the occluded part of the ring-shaped part, according to the principle of "drawing an arc through three points", two endpoints of the existing profile and a point on the profile close to the endpoints are selected to draw an arc, and the part that coincides with the existing profile is cut off to obtain the profile at the occlusion.

[0013] Further, the specific steps include:

[0014] (1) Preparation work: Place the initial blank in the processing area of the ring rolling equipment, and install an industrial camera directly above it to be responsible for obtaining the top view, and install an industrial camera on the side to be responsible for obtaining the front view;

[0015] (2) Start-up and operation: Connect the power supply, the ring rolling equipment starts to process the blank, the industrial camera starts to obtain RGB images, and inputs the RGB image information to the communication computer;

[0016] (3) Depth map prediction: Taking the RGB image obtained in step (2) as the input, through the depth prediction model, the corresponding depth image is predicted;

[0017] (4) Semantic segmentation of key regions: Taking the depth map obtained in step (3) as the input, through the lightweight semantic segmentation model, the masks of key regions such as the ring-shaped part, the cone roller, and the core roller are predicted;

[0018] (5) Extract the ring-shaped part region: According to the ring-shaped part mask obtained in step (4), the RGB image obtained in step (2) is segmented to obtain the RGB information of the ring-shaped part region;

[0019] (6) Filtering processing: Perform filtering processing on the RGB information of the ring-shaped part region obtained in step (5) to remove color noise and achieve a smooth effect;

[0020] (7) Edge detection: Perform edge detection on the RGB information processed in step (6). Set the hyperparameters of the edge detection algorithm according to the actual working conditions to obtain clear edge information;

[0021] (8) Useless edge clipping: According to the masks of the cone roller and the core roller obtained in step (4), perform cutting processing on the edge information obtained in step (7), that is, remove the contour information of the cone roller and the core roller;

[0022] (9) Inner and outer ring area separation: Through a clustering algorithm, classify the edge information obtained in step (8) to obtain the edge information of the inner ring area and the outer ring contour respectively;

[0023] (10) Inner ring area contour refinement: Set a reasonable depth value threshold. According to the depth information obtained in step (3), screen the edge information of the inner ring area obtained in step (9) to achieve the accurate extraction of the inner ring contour of the upper bottom surface;

[0024] (11) Inner and outer ring contour completion: Process the outer ring obtained in step (9) and the inner ring obtained in step (10) respectively to complete the contour;

[0025] (12) Annular part height acquisition: If the height of the annular part remains unchanged during the processing, the initial value of the height is adopted. If the height of the annular part changes, the height of the annular part is obtained using the principles of steps (3) to (4);

[0026] (13) Geometric 3D reconstruction of the annular part: According to the complete inner and outer ring contours obtained in step (11) and the height of the annular part obtained in step (12), use the existing 3D modeling method to perform rapid 3D reconstruction of the annular part;

[0027] (14) Repeat steps (3) to (12) for each received frame of RGB image. Each repetition takes about 80 ms to meet the real-time requirement;

[0028] (15) Process end: Wait until the annular part reaches the required size and stop the entire process.

[0029] Furthermore, the initial blank in step (1) includes circular, oval, triangular, rectangular or custom shapes; and if the height of the annular part remains unchanged during the ring rolling process, there is no need to install an industrial camera on the side to obtain a front view.

[0030] Furthermore, the depth prediction model in step (3) is the Depth Anything v2 model.

[0031] Furthermore, the lightweight semantic segmentation model in step (4) is the Mobile SAM model.

[0032] It should be noted that step (3) of the present invention provides rich depth information for the contour extraction of the annular part, which can effectively help to eliminate the useless contours that do not conform to the depth value range, thereby improving the accuracy of the annular part contour extraction. Moreover, compared with using the RGB image as the input data for step (4), using the depth image obtained in step (3) as the input data for step (4) can effectively avoid the problem of reduced semantic segmentation accuracy caused by too high color information, thereby significantly improving the segmentation accuracy of the annular part, the tapered roller and the core roller, and further improving the accuracy of the annular part contour extraction. Therefore, based on advanced computer vision technology, the present invention makes full use of the depth prediction model and semantic segmentation model that do not require additional training, and combines image processing algorithms to realize the real-time extraction of the contours of annular parts with various bottom shapes.

[0033] Furthermore, the hyperparameters of the edge detection algorithm in step (7) are the minimum threshold threshold1 and the maximum threshold threshold2 of the Canny algorithm.

[0034] Furthermore, the clustering algorithm in step (9) includes the DBSCAN algorithm, the Spectral Clustering algorithm or the OPTICS algorithm.

[0035] Furthermore, the method for contour completion in step (11) is as follows: according to the principle of "drawing an arc through three points", select two endpoints of the existing contour and a point on the contour close to the endpoint to draw an arc, and cut off the part that coincides with the existing contour to obtain the contour of the occluded part; or if it is determined that the three points are on the same straight line, then use the method of "directly connecting the three points" to complete the contour of the occluded part.

[0036] Therefore, the present invention uses an industrial camera as an image data acquisition device, and through means such as camera pose optimization and data communication, it can perceive and obtain the geometric morphology information (RGB image) of the ring-shaped part in the intelligent ring forging process in real time. Taking the RGB image as the input data, without additional training and learning, directly through the depth prediction model Depth Anything v2, a depth image that meets the requirements can be obtained in real time. Input the depth image and semantic prompt points into the semantic segmentation model Mobile SAM to obtain the mask of the ring-shaped part (i.e., Mask information) in real time. Based on the mask result, morphological operations and contour recognition are carried out, and through the clustering algorithm, the contour points of the inner and outer rings of the upper bottom surface of the ring-shaped part with high accuracy can be quickly obtained. According to the same principle, taking the front view of the ring part as the input, the height of the ring part can be obtained, or when ensuring that the height of the ring part remains unchanged, the height of the ring part can be directly obtained according to the initial parameters. For the part of the ring-shaped part blocked by the device, according to the principle of "drawing an arc with three points", select two endpoints of the existing contour and a point on the contour close to the endpoint (referred to as the "third contour point") to draw an arc, and cut off the part that coincides with the existing contour to obtain the contour at the occlusion. Finally, through algorithm optimization, on the premise of ensuring accuracy, the real-time extraction speed of the ring-shaped part contour of about 15 frames per second is achieved, and the geometric three-dimensional reconstruction of the ring part is quickly completed according to the contour points of the upper bottom surface of the ring-shaped part and the height of the ring-shaped part. Figure 1 )。

[0037] In summary, the present invention can meet the requirements of real-time contour extraction and rapid geometric three-dimensional reconstruction, and can complement the missing part of the contour. The entire computer vision algorithm route is suitable for ring-shaped parts with various bottom shapes and has good robustness. Compared with existing technologies such as laser scanning measurement and multi-vision stereo measurement, the present invention can extract the contour of the ring-shaped part faster on the premise of ensuring accuracy, meeting the requirements of online extraction and measurement. At the same time, the entire implementation route of the present invention has strong robustness, does not require multiple scans, and can obtain the complete RGB image data at one time, is suitable for ring-shaped parts with various bottom shapes, and has a wider application scenario. The bottom shape of the ring-shaped part includes but is not limited to circular, elliptical, triangular, rectangular, and other custom shapes. Figure 2 )。And the present invention uses an industrial camera and a computer as implementation devices, which have lower costs and are more conducive to industrial practice compared with laser scanning devices or multi-vision stereo imaging devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the provided drawings.

[0039] Figure 1 This is the algorithm flow for extracting the contour of the annular part in each frame of RGB image of the present invention.

[0040] Figure 2 These are some annular parts applicable to the algorithm route of the present invention.

[0041] Figure 3 This is the schematic diagram of the algorithm route for preliminary contour extraction in Embodiment 1 of the present invention.

[0042] Figure 4 This is the schematic diagram of the algorithm route for contour clustering and completion in Embodiment 1 of the present invention.

[0043] Figure 5 This is the result of the geometric three-dimensional reconstruction according to the extracted contour and height in Embodiment 1 of the present invention. Detailed implementation manners

[0044] Next, the technical solutions in the embodiments of the present invention will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0045] Here, the special term "embodiment", any embodiment described as "exemplary" does not have to be construed as superior to or better than other embodiments. For the performance index tests in the embodiments of this application, unless otherwise specified, the conventional test methods in the art are adopted. It should be understood that the terms described in this application are only used to describe specific implementation manners and are not used to limit the content disclosed in this application.

[0046] Unless otherwise specified, the technical and scientific terms used in this article have the same meanings as those generally understood by those of ordinary skill in the technical field to which this application belongs; the test methods and technical means not specifically mentioned in other parts of this application refer to the experimental methods and technical means commonly adopted by those of ordinary skill in the art.

[0047] To better illustrate the content of this application, numerous specific details are given in the following specific embodiments. Those skilled in the art should understand that this application can still be implemented without some specific details. In the embodiments, some methods, means, instruments, equipment, etc. well-known to those skilled in the art are not described in detail to highlight the gist of this application.

[0048] On the premise of no conflict, the technical features disclosed in the embodiments of this application can be combined arbitrarily, and the obtained technical solutions belong to the content disclosed in the embodiments of this application.

[0049] The present invention belongs to the field of optoelectronic measurement technology, and particularly relates to a method for real-time extraction and three-dimensional reconstruction of the contour of an intelligent ring rolling formed ring forging. Through the idea of model cascading, the present invention inputs the predicted depth map into a semantic segmentation model, then performs edge detection algorithms and morphological operations on the segmentation results, and uses a contour completion algorithm to achieve the on-line contour extraction of the ring-shaped part during the intelligent ring forging process, and then realizes the reconstruction of the three-dimensional geometric model. It is applicable to ring forgings with various bottom shapes and has good robustness. In addition, this method can effectively complete the missing part of the contour caused by equipment occlusion, ensuring the integrity and accuracy of the contour data.

[0050] To better understand the present invention, the following further specifically elaborates the present invention through the following embodiments. However, it should not be understood as a limitation of the present invention. For some non-essential improvements and adjustments made by those skilled in the art based on the above invention content, they are also considered to fall within the protection scope of the present invention.

[0051] Embodiment 1

[0052] A method for real-time extraction and three-dimensional reconstruction of the contour of an intelligent ring rolling formed ring forging:

[0053] (1) Preparation work: Place the initial blank with a circular bottom shape in the processing area of the ring rolling equipment, and install an industrial camera directly above it to be responsible for obtaining the top view.

[0054] (2) Start-up and operation: Connect the power supply, the ring rolling equipment starts to process the blank, the industrial camera starts to obtain RGB images, and inputs the RGB images into the communication computer.

[0055] (3) Depth map prediction: Taking the RGB image obtained in step (2) as the input, through the Depth Anything v2 model, the corresponding depth image is predicted.

[0056] (4) Semantic segmentation of key regions: Taking the depth map obtained in step (3) as the input, through the Mobile SAM model, the masks of key regions such as the ring-shaped part, conical roller, and core roller are predicted.

[0057] (5) Extract the ring-shaped part region: According to the ring-shaped part mask obtained in step (4), segment the RGB image obtained in step (2) to obtain the RGB information of the ring-shaped part region.

[0058] (6) Filtering processing: Perform filtering processing on the RGB information of the ring-shaped part region obtained in step (5) to remove color noise and achieve a smooth effect.

[0059] (7) Edge detection: Apply the Canny algorithm to the RGB information processed in step (6). Set the minimum threshold threshold1 of the Canny algorithm to 30 and the maximum threshold threshold2 to 180 to obtain clear edge information;

[0060] (8) Useless edge cropping: According to the masks of the tapered roller and the core roller obtained in step (4), perform cutting processing on the edge information obtained in step (7), that is, remove the contour information of the tapered roller and the core roller;

[0061] (9) Inner and outer ring area separation: Apply the DBSCAN algorithm to classify the edge information obtained in step (8) to obtain the edge information and the outer ring contour of the inner ring area respectively;

[0062] (10) Inner ring area contour refinement: Set the depth value threshold to 105. According to the depth information obtained in step (3), screen the edge information of the inner ring area obtained in step (9) to achieve the accurate extraction of the inner ring contour of the upper bottom surface;

[0063] (11) Inner and outer ring contour completion: Process the outer ring obtained in step (9) and the inner ring obtained in step (10) respectively. Select to use the "three-point arc drawing" algorithm. Draw an arc through two endpoints of the existing contour and a point on the contour close to the endpoint (the interval number from the third contour point to the endpoint is set to 50), and cut off the part that coincides with the existing contour to obtain the contour of the occluded part;

[0064] (12) Obtaining the height of the ring part: During this processing, the height of the ring part remains unchanged at 9 cm;

[0065] (13) Geometric three-dimensional reconstruction of the ring part: According to the complete inner and outer ring contours and the height of the ring part obtained in step (11), use the python library to perform rapid three-dimensional reconstruction of the ring part;

[0066] (14) Repeat steps (3) to (12) for each received RGB image. Each repetition takes about 78 ms, meeting the real-time requirement;

[0067] (15) Process end: When the inner diameter of the ring part reaches 70 cm and the outer diameter reaches 90 cm ( Figure 5 ), stop the entire process.

[0068] The foregoing description of the disclosed embodiments enables those skilled in the art to practice or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for real-time contour extraction and three-dimensional reconstruction of an intelligent rolling-formed ring forging, characterized in that: An industrial camera is used as an image data acquisition device. Through camera posture optimization and data communication, the geometric morphology information RGB image of the ring part in the intelligent ring forging process is perceived in real time and obtained. Then, the RGB image is used as input data to obtain the depth image in real time through the depth prediction model. The depth image and semantic cue points are input into the semantic segmentation model to obtain the mask of the ring in real time; morphological operations and contour recognition are performed based on the mask results, and the contour points of the inner and outer rings of the upper and lower surfaces of the ring are obtained through a clustering algorithm; and the height of the ring is obtained using the front view of the ring as input; finally, the geometric 3D reconstruction of the ring is completed based on the contour points of the upper and lower surfaces of the ring and the height of the ring through algorithm optimization.

2. The method according to claim 1, characterized in that When the height of the ring member remains unchanged, the height of the ring member is obtained according to the initial parameters.

3. The method according to claim 1, characterized in that For the blocked part of the ring, according to the principle of "drawing an arc from three points", select the two end points of the existing contour and a point on the contour close to the end points to draw an arc, cut off the part overlapping with the existing contour, and obtain the contour of the blocked part.

4. The method according to any one of claims 1 to 3, characterized in that: The specific steps include: (1) Preparatory work: Place the initial blank in the processing area of ​​the ring rolling equipment, place an industrial camera directly above it to obtain a top view, and place an industrial camera on the side to obtain a front view; (2) Start-up: Turn on the power, the ring rolling equipment starts to process the blank, the industrial camera starts to acquire RGB images, and inputs RGB image information to the communication computer; (3) Depth map prediction: Using the RGB image obtained in step (2) as input, the corresponding depth image is predicted through the depth prediction model; (4) Semantic segmentation of key areas: Using the depth map obtained in step (3) as input, a lightweight semantic segmentation model is used to predict masks of key areas such as ring parts, cone rollers, and core rollers. (5) Extracting the annular region: segmenting the RGB image obtained in step (2) according to the annular mask obtained in step (4) to obtain RGB information of the annular region; (6) filtering: filtering the RGB information of the annular area obtained in step (5) to remove color noise and achieve a smooth effect; (7) Edge detection: Perform edge detection on the RGB information processed in step (6), and set the hyper parameters of the edge detection algorithm according to the actual working conditions to obtain clear edge information; (8) useless edge cutting: cutting the edge information obtained in step (7) according to the mask of the cone roller and the core roller obtained in step (4), that is, removing the contour information of the cone roller and the core roller; (9) Separation of inner and outer ring areas: Using a clustering algorithm, the edge information obtained in step (8) is classified to obtain the edge information of the inner ring area and the outer ring contour respectively; (10) Refining the inner ring contour: Setting a reasonable depth value threshold, filtering the inner ring edge information obtained in step (9) based on the depth information obtained in step (3), and realizing accurate extraction of the inner ring contour of the upper bottom surface; (11) inner and outer ring contour completion: respectively process the outer ring obtained in step (9) and the inner ring obtained in step (10) to complete the contours; (12) Obtaining the height of the ring part: If the height of the ring part remains unchanged during the processing, the initial value of the height is used; if the height of the ring part changes, the height of the ring part is obtained by using the principles of steps (3) to (4); (13) 3D reconstruction of the ring geometry: based on the complete inner and outer ring contours obtained in step (11) and the height of the ring obtained in step (12), the existing 3D modeling method is used to perform rapid 3D reconstruction of the ring; (14) Repeat (3) to (12) for each frame of RGB image received. Each repetition takes about 80 ms, meeting the real-time requirement. (15) Process end: When the ring part reaches the required size, the entire process is stopped.

5. The method according to claim 4, characterized in that In the step (1), the initial blank includes a circular, oval, triangular, rectangular or custom shape; and if the height of the ring part remains unchanged during the rolling forming process, there is no need to place an industrial camera on the side to obtain a front view.

6. The method according to claim 4, characterized in that The depth prediction model in step (3) is the DepthAnything v2 model.

7. The method according to claim 4, characterized in that The lightweight semantic segmentation model in step (4) is the Mobile SAM model.

8. The method according to claim 4, characterized in that The hyper parameters of the edge detection algorithm in step (7) are the minimum threshold value threshold1 and the maximum threshold value threshold2 of the Canny algorithm.

9. The method according to claim 4, characterized in that The clustering algorithm in step (9) includes DBSCAN algorithm, Spectral Clustering algorithm or OPTICS algorithm.

10. The method according to claim 4, characterized in that The method for completing the contour in step (11) is as follows: based on the principle of "drawing an arc from three points", two endpoints of the existing contour and a point on the contour close to the endpoints are selected to draw an arc, and the part overlapping with the existing contour is cut off to obtain the contour of the blocked part; or if it is determined that the three points are on the same straight line, the "direct connection of the three points" method is used to complete the contour of the blocked part.

Images