A method and system for linear feature recognition of structural components based on multi-order segmentation

Abstract

The application provides a kind of structure component linear fast identification method and system based on multi-order segmentation, related to engineering structure intelligent detection technical field, method includes: obtaining the original image of structure component;The original image is roughly segmented, and the initial binary mask of structure component is generated;By connected domain analysis mode, the target mask region containing foreground prompt point is automatically screened out;By contour detection mode, the circumscribed rectangle boundary box coordinate information of target mask region is calculated;The corresponding target region sub-image is extracted from the original image;Using super-resolution reconstruction model, the target region sub-image is enhanced;The subgraph of the target region sub-image after enhancement is divided;The subgraph is finely segmented, and high-precision mask is determined;Through coordinate mapping, high-precision mask is mapped to the original image;Combined with edge detection algorithm, the outer contour line of structure component is extracted, and the structure component linear identification is completed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent detection of engineering structures, and in particular to a structure member linear rapid identification method and system based on multi-order segmentation. BACKGROUND

[0002] The structure member linear rapid identification method based on multi-order segmentation is a new type of identification method that combines computer vision and image enhancement technology, mainly used for efficiently and accurately extracting the linear profile of the structure member in the engineering structure from the image, better restoring the edge details of the structure member in low-quality or complex background images, reducing human intervention and data dependence, and having excellent precision, robustness and engineering practicability, and being suitable for structure health monitoring and intelligent identification tasks in multiple scenes such as buildings, bridges and tunnels.

[0003] In the field of buildings and infrastructure, the linear profile of the structure member is a key geometric element representing the spatial form, stress characteristics and connection relationship of the structure, and its rapid and accurate identification is of great significance for ensuring structure safety, improving construction precision, realizing quality traceability, and carrying out structure health monitoring and deformation warning in the operation stage.

[0004] However, the traditional structure member linear rapid identification method usually uses total station measuring points or three-dimensional laser scanning equipment, which has the defects of high cost, complex data processing, poor real-time performance, and inability to cope with large-scale and high-frequency monitoring tasks, and is easily affected by environmental interference and human error. The traditional method introduces a deep learning algorithm for rapid identification, but this algorithm is heavily dependent on large-scale high-quality labeled data, has high training cost, and the boundary recognition is not clear in low-resolution and fuzzy detail scenes, which leads to problems such as missegmentation and boundary drift in the edge region of the structure member, affecting stability and generalization ability. SUMMARY

[0005] In view of the above deficiencies of the prior art, the purpose of the embodiments of the present application is to provide a structure member linear rapid identification method based on multi-order segmentation, which can solve the technical problems that the traditional structure member linear rapid identification method usually uses total station measuring points or three-dimensional laser scanning equipment, which has the defects of high cost, complex data processing, poor real-time performance, and inability to cope with large-scale and high-frequency monitoring tasks, and is easily affected by environmental interference and human error. The traditional method introduces a deep learning algorithm for rapid identification, but this algorithm is heavily dependent on large-scale high-quality labeled data, has high training cost, and the boundary recognition is not clear in low-resolution and fuzzy detail scenes, which leads to problems such as missegmentation and boundary drift in the edge region of the structure member, affecting stability and generalization ability.

[0006] The first aspect of the embodiment of the present application provides a structure member linear fast identification method based on multi-order segmentation, comprising:

[0007] S1: obtaining an original image of a structure member;

[0008] S2: performing rough segmentation on the original image by using a first visual segmentation large model to generate an initial binary mask of the structure member;

[0009] S3: based on a foreground prompt point in the original image, automatically screening out a target mask region containing the foreground prompt point in the initial binary mask by a connected domain analysis method;

[0010] S4: calculating the coordinates of a circumscribed rectangle bounding box of the target mask region by a contour detection method;

[0011] S5: extracting a corresponding target region sub-image from the original image according to the coordinates of the rectangle bounding box;

[0012] S6: enhancing the target region sub-image by using a super-resolution reconstruction model;

[0013] S7: performing subgraph division on the enhanced target region sub-image to determine a plurality of subgraphs;

[0014] S8: performing fine segmentation on each of the subgraphs by a second visual segmentation large model to determine a high-precision mask;

[0015] S9: mapping the high-precision mask to the original image by coordinate mapping;

[0016] S10: extracting an outer contour line of the structure member by combining an edge detection algorithm to complete structure member linear identification.

[0017] The second aspect of the embodiment of the present application provides a structure member linear fast identification system based on multi-order segmentation, comprising: a processor and a memory.

[0018] The memory stores programs or instructions executable on the processor, and the programs or instructions are executed by the processor to implement the steps of the structure member linear fast identification method based on multi-order segmentation as described in the first aspect.

[0019] The third aspect of the embodiment of the present application provides a readable storage medium, and the readable storage medium stores programs or instructions, and the programs or instructions are executed by a processor to implement the steps of the structure member linear fast identification method based on multi-order segmentation as described in the first aspect.

[0020] The technical scheme provided by the embodiment of the present application has at least the following beneficial effects:

[0021] In the embodiment of the present application, a point prompt supporting visual segmentation large model is used for rough segmentation, automatic candidate region screening is completed, manual annotation dependence is reduced, the automation level is improved, the rapid positioning of the target region is realized under the condition of no training, the image resolution, edge definition and texture details are significantly improved through the introduction of super-resolution reconstruction, the subsequent segmentation module can still maintain high recognition accuracy when processing blurred, compressed or long-distance shooting images, effectively breaking through the low-quality image bottleneck, the sliding window mechanism is adopted to realize accurate identification at the sub-image level, especially in the boundary area, the recognition accuracy and edge definition of the structure member contour in the complex background and low-quality image conditions are significantly improved by combining the super-resolution image enhancement and multi-stage fine segmentation strategy. BRIEF DESCRIPTION OF DRAWINGS

[0022] The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings:

[0023] Figure 1 is a flowchart of a structure member linear rapid identification method based on multi-order segmentation provided by an embodiment of the present application;

[0024] Figure 2 is a flowchart of a structure member linear rapid identification method based on multi-order segmentation provided by an embodiment of the present application;

[0025] Figure 3 is a structure diagram of a structure member linear rapid identification system based on multi-order segmentation provided by an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order for those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions of the present application will be described clearly and completely below in conjunction with the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0027] The structure member linear rapid identification method based on multi-order segmentation provided by the embodiments of the present application will be described in detail below in conjunction with the drawings through specific embodiments and application scenarios.

[0028] Referring to the accompanying drawings Figure 1 , a flowchart of a structure member linear fast identification method based on multi-stage segmentation provided by an embodiment of the present application is shown.

[0029] Referring to the accompanying drawings Figure 2 , a flowchart of a structure member linear fast identification method based on multi-stage segmentation provided by an embodiment of the present application is shown.

[0030] As Figure 2 shown, in the present application, first, by inputting an image, a target contour based on foreground point coordinates is obtained by using a support point prompt and an interactive segmentation model for preliminary processing. Then, by an edge extraction method, the target boundary is further extracted and optimized. Then, combined with structured target information, detailed processing is performed to extract the detailed contour of the target. Finally, super-resolution technology and sliding window algorithm are used to accurately repair and extract the target, and finally multiple refined results are generated to complete the accurate identification and extraction of the target.

[0031] An embodiment of the present application provides a structure member linear fast identification method based on multi-stage segmentation, which can include the following steps:

[0032] S1: obtaining an original image of a structure member.

[0033] Wherein, the structure member refers to important components in building, bridge, tunnel and other engineering structures that bear load function or constitute the whole structure.

[0034] S2: using a first visual segmentation large model to perform coarse segmentation on the original image to generate an initial binary mask of the structure member.

[0035] Wherein, the visual segmentation large model is preferably a general segmentation model with support point prompt interaction, such as SegmentAnything Model (SAM) or its lightweight variant, which can quickly locate the target area and generate a rough mask according to user or automatically generated point prompts. This model has strong cross-scene generalization ability and strong adaptability, and can complete initial target positioning without fine-tuning for specific structure members, thereby improving the automation level and migratability of the entire segmentation process.

[0036] It should be noted that by introducing the first visual segmentation large model, the target area can be quickly located under unsupervised or weakly supervised conditions, and the initial mask of the structure member is generated to provide a stable and efficient input basis for subsequent multi-stage refinement processing, effectively improving the system automation level and cross-scene adaptability.

[0037] In a possible implementation, the first visual segmentation large model is specifically a general visual segmentation large model supporting point prompt interaction.

[0038] For example, the general visual segmentation large model supporting point prompt interaction can be SAM, RITM, FocalClick, or CLIPSeg.

[0039] S3: Based on the foreground prompt point in the original image, a target mask region containing the foreground prompt point is automatically screened out in the initial binary mask through a connected component analysis method.

[0040] The foreground prompt point refers to a feature point in the image that is selected by a user or a model to identify a target region. These points are usually located inside the target region of interest and are used to guide the model to identify the location of the target region. Connected component analysis is a technique used in image processing to detect continuous pixel regions in an image and identify "connected" regions. Specifically, it groups all regions in the image that have the same pixel value and are adjacent to each other.

[0041] It should be noted that by combining the foreground prompt point and the connected component analysis, the region containing the target can be accurately screened out, thereby avoiding missegmentation or missed segmentation, enhancing the accuracy and reliability of mask screening, and providing more accurate input for subsequent fine processing.

[0042] In a possible implementation, S3 is specifically:

[0043] Based on the determination principle that the foreground prompt point coordinates are located inside the target mask region, a target mask region containing the foreground prompt point is automatically screened out in the initial binary mask through a connected component analysis method.

[0044] Specifically, the determination principle that the foreground prompt point coordinates are located inside the target mask region is specifically: if a "mask region" contains foreground prompt point coordinates, it is determined that the region is a real target region, otherwise, the region is considered to be a missegmentation region and should be excluded.

[0045] In the embodiments of the present application, the connected component analysis method is specifically: performing morphological operations on the binary mask image, and then applying eight-neighbor or four-neighbor connection rules to extract all connected regions. The region containing the foreground prompt point coordinates is automatically screened out as the target mask region, and a circumscribed rectangle frame is generated according to the boundary contour of the region for target positioning.

[0046] S4: Calculate the circumscribed rectangular boundary box coordinate information of the target mask region through a contour detection method.

[0047] Among them, the contour detection is an image processing method for identifying the edge boundary with the same pixel value in the image, that is, determining the "shape" or "edge line" of the target region, and the contour detection method is usually findContours() function in OpenCV.

[0048] It should be noted that the target region boundary is extracted by the contour detection, and the bounding rectangle boundary box coordinates are calculated, so that the accurate positioning of the spatial position of the structural member can be realized, accurate reference boundary is provided for subsequent image cropping and region enhancement, and the positioning accuracy and processing efficiency of the overall process are significantly improved.

[0049] S5: Extracting the corresponding target region sub-image from the original image according to the rectangular boundary box coordinates.

[0050] It should be noted that the target region sub-image is extracted from the original image according to the boundary box coordinates, which can effectively focus on the region of interest, reduce background interference and computational burden, and provide clear and accurately positioned input blocks for subsequent image enhancement and fine segmentation, thereby significantly improving the overall processing efficiency and recognition accuracy.

[0051] In the embodiment of the application, the extracted target region sub-image can be set to be slightly larger than the bounding rectangle in size, so as to avoid the loss of member boundary information due to truncation and enhance the robustness.

[0052] S6: Enhancing the target region sub-image by using a super-resolution reconstruction model.

[0053] The super-resolution reconstruction model is an image enhancement technology based on deep learning, which aims to reconstruct a low-resolution image into a high-resolution image to restore details, edges and texture information.

[0054] It should be noted that the target region sub-image is enhanced by using the super-resolution reconstruction model, which can significantly improve the clarity and edge texture quality of the image, so that the subsequent segmentation model is more accurate in identifying the member contour and detail boundary, especially when processing blurred, compressed or long-distance collected images.

[0055] In one possible implementation, the super-resolution reconstruction model is specifically a reconstruction model constructed based on a deep convolutional neural network.

[0056] The reconstruction model constructed based on the deep convolutional neural network can be Real-ESRGAN, EDSR and RCAN, which can effectively improve the image edge quality while preserving the image texture details, and facilitate the accurate judgment of the complex boundary by the subsequent segmentation model.

[0057] S7: Dividing the enhanced target region sub-image into subgraphs to determine a plurality of subgraphs.

[0058] It should be noted that by dividing the enhanced target region image into multiple subgraphs, not only the modeling capability of the local region is improved, but also the recognition accuracy of the edge, detail and small component is effectively improved, and the parallel segmentation processing at the subgraph level is facilitated, and the overall calculation efficiency and scalability are improved.

[0059] In a possible implementation, S7 specifically is:

[0060] Through the sliding window mechanism, the enhanced target region subgraph image is divided into multiple non-overlapping or optionally overlapping subgraphs.

[0061] In the embodiment of the application, the window size and step size of the sliding window can be flexibly adjusted according to the size of the target region image and the segmentation requirement. The non-overlapping strategy can realize complete coverage of the region by separately dividing the boundary region into a supplementary window, and in a specific implementation, it can also be switched to an overlapping sliding window strategy to optimize the segmentation consistency of the boundary transition region and improve the fusion quality.

[0062] S8: performing fine segmentation on each subgraph by using a second visual segmentation large model to determine a high-precision mask.

[0063] It should be noted that by using the second visual segmentation large model to perform fine segmentation on the subgraph, the segmentation accuracy of the boundary contour can be greatly improved while retaining the structural component details, which is particularly suitable for processing regions with complex shapes or blurred edges, and lays a solid foundation for high-quality contour extraction.

[0064] In a possible implementation, the second visual segmentation large model specifically is a visual segmentation large model guided by a prompt point and a rectangular box.

[0065] In the embodiment of the application, the visual segmentation large model guided by the prompt point and the rectangular box specifically can be a SAM model, a SAM2 model, a FastSAM model and a SEEM model.

[0066] S9: mapping the high-precision mask to the original image through coordinate mapping.

[0067] It should be noted that by mapping the high-precision mask back to the original image coordinate system, the consistency of the segmentation result and the original image is ensured, which facilitates subsequent visualization and further analysis, and improves the accuracy and practicality of the overall processing flow.

[0068] In a possible implementation, S9 specifically includes:

[0069] S901: mapping the high-precision mask to the local coordinate system of the target region subgraph image through coordinate mapping.

[0070] S902: Map the high-precision mask to the original image according to the coordinate position of the target region sub-image in the original image, and extract the outer contour line of the structural member in combination with the edge detection algorithm.

[0071] S10: Extract the outer contour line of the structural member in combination with the edge detection algorithm to complete the structural member line type identification.

[0072] The edge detection algorithm is an image processing technology for identifying regions in an image where there is a significant change in brightness or color, i.e., the edges of the image.

[0073] It should be noted that by extracting the outer contour line of the structural member in combination with the edge detection algorithm, the boundary and shape of the member can be accurately identified, providing accurate basic data for subsequent structural analysis, modeling and visualization, and improving the accuracy and efficiency of the overall identification.

[0074] In one possible implementation, the edge detection method is specifically Canny algorithm, Sobel algorithm or HED algorithm driven by deep learning.

[0075] In the embodiments of the present application, the edge detection method is preferably Canny, Sobel or HED (Holistically-Nested Edge Detection) algorithm driven by deep learning, which detects the mask boundary and extracts the pixel point coordinates on the boundary contour, connects the adjacent coordinates to finally generate the line type contour graph of the structural member, and can be rendered in pseudo-color to realize visual output.

[0076] The technical solutions provided by the embodiments of the present application have at least the following beneficial effects:

[0077] In the embodiments of the present application, a visual segmentation large model supporting point hints is used for coarse segmentation to automatically complete candidate region screening, reduce the dependence on manual annotation, improve the automation level, realize rapid positioning of the target region without training, and significantly improve the image resolution, edge definition and texture details by introducing super-resolution reconstruction, so that the subsequent segmentation module can still maintain high recognition accuracy when processing blurred, compressed or remotely shot images, effectively breaking through the bottleneck of low-quality images, and adopting a sliding window mechanism combined with adaptive fine segmentation scheme of local hint points to realize accurate identification at the sub-image level, especially in the boundary area. The present application combines super-resolution image enhancement and multi-stage fine segmentation strategy to significantly improve the recognition accuracy and edge definition of the structural member contour in complex background and low-quality image conditions.

[0078] Reference is made to the accompanying drawings Figure 3 , which shows a structural schematic diagram of a structural member line type rapid identification system based on multi-stage segmentation provided by an embodiment of the present application.

[0079] The embodiment of the present application provides a structure member linear fast identification system 20 based on multi-order segmentation, which comprises a processor 201 and a memory 202.

[0080] The memory 202 stores programs or instructions which can be run on the processor 201, and the programs or instructions are executed by the processor 201 to realize the steps of the above-mentioned structure member linear fast identification method based on multi-order segmentation, and the same technical effects can be achieved. To avoid repetition, the present application will not be described again.

[0081] It should be understood that the processor 201 in the embodiment of the present application can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), ready programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0082] It is also to be understood that the memory 202 in embodiments of the present application can be volatile or nonvolatile memory, or can include both volatile and nonvolatile memory. Nonvolatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrically EPROM (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), used as external cache. By way of example, and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0083] The above-described embodiments can be implemented in whole or in part by software, hardware (e.g., circuitry), firmware, or any combination of the three. When implemented in software, the above-described embodiments can be implemented in the form of one or more computer programs that are stored in a computer-readable storage medium. The computer-readable storage medium stores one or more computer instructions or computer programs that, when loaded into a computer, cause the computer to perform the processes or functions described in the embodiments of the present application. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable apparatus. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, such as from a website, a computer, a server, or a data center to another website, computer, server, or data center, via a wired (e.g., infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more collections of available media. The available media can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0084] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0085] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those of ordinary skill in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0086] Those of ordinary skill in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the devices, apparatuses and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0087] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other manners. For example, the embodiments of the apparatus described above are merely schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation. For example, a plurality of units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0088] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0089] In addition, each functional unit in the various embodiments of the present application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0090] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the present application that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0091] The embodiment of the present application provides a readable storage medium, which includes: a program or instructions stored on the readable storage medium, the program or instructions are executed by a processor to realize the steps of the above-mentioned structure member linear fast identification method based on multi-order segmentation, and the same technical effects can be achieved. To avoid repetition, the present application will not be described again.

[0092] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present application, but not to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application. Any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered within the protection scope of the present application.

Claims

1. A method for fast recognition of linear structures in a component based on multi-order segmentation, characterized in that, The method comprises the following steps: S1: obtaining an original image of a structural member; S2: performing rough segmentation on the original image by using a first visual segmentation large model to generate an initial binary mask of the structural member; wherein the first visual segmentation large model is specifically a general visual segmentation large model supporting point prompt interaction; S3: based on a foreground prompt point in the original image, automatically screening out a target mask region containing the foreground prompt point in the initial binary mask through a connected domain analysis method; S4: calculating the coordinates of the bounding box of the circumscribed rectangle of the target mask region through a contour detection method; S5: extracting a corresponding target region sub-image from the original image according to the coordinates of the rectangular bounding box; S6: enhancing the target region sub-image by using a super-resolution reconstruction model; S7: dividing the enhanced target region sub-image into multiple sub-images to determine the multiple sub-images; wherein S7 is specifically: dividing the enhanced target region sub-image into multiple non-overlapping or optionally overlapping sub-images through a sliding window mechanism to determine the multiple sub-images; S8: performing fine segmentation on each of the sub-images by using a second visual segmentation large model to determine a high-precision mask; wherein the second visual segmentation large model is specifically a visual segmentation large model guided by the prompt point and the rectangular box; S9: mapping the high-precision mask to the original image through coordinate mapping; S10: extracting the outer contour line of the structural member by combining an edge detection algorithm to complete the line type recognition of the structural member.

2. The method of claim 1, wherein the method is a multi-stage segmentation based structural member line type quick identification method. S3 is specifically: based on the determination principle that the foreground prompt point coordinates are located inside the target mask region, the target mask region containing the foreground prompt point is automatically screened out in the initial binary mask through the connected domain analysis method.

3. The method of claim 1, wherein the method is a multi-stage segmentation based structural member line type quick identification method. The super-resolution reconstruction model is specifically a reconstruction model constructed based on a deep convolutional neural network.

4. The method of claim 1, wherein the method is a multi-stage segmentation based structural member line type quick identification method. S9 specifically includes: S901: mapping the high-precision mask to the local coordinate system of the target region sub-image through coordinate mapping; S902: mapping the high-precision mask to the original image according to the coordinate position of the target region sub-image in the original image, and extracting the outer contour line of the structural member by combining an edge detection algorithm.

5. The method of claim 1, wherein the method is a multi-stage segmentation based structural member line type quick identification method. The edge detection method is specifically: Canny algorithm, Sobel algorithm or deep learning driven HED algorithm.

6. A multi-order segmentation based structural member linear quick identification system, characterized in that, The method comprises: a processor and a memory; the memory stores programs or instructions that can be run on the processor, and the programs or instructions are executed by the processor to realize the steps of the multi-stage segmentation-based structural member line type fast recognition method according to any one of claims 1 to 5.

7. A readable storage medium, characterized by, The programs or instructions are stored on the readable storage medium, and the programs or instructions are executed by the processor to realize the steps of the multi-stage segmentation-based structural member line type fast recognition method according to any one of claims 1 to 5.

Images