Vehicle damage state identification and analysis method and system based on multi-modal data

Abstract

The invention relates to the technical field of vehicle scratch detection, in particular to a vehicle damage state recognition and analysis method and system based on multi-modal data, and the method comprises the steps: carrying out the real-time monitoring of a target vehicle, and immediately generating a detection signal when a designated contact object is recognized; obtaining a front feature image and a termination image, carrying out difference processing to obtain a difference image, then identifying the type of a specified contact object, counting the number of damage occurrence times and the total number of contact times, calculating the damage probability, and comparing the difference image with scar features to determine a scar similarity value; and finally, performing weighting processing on the damage probability and the scar similarity value to obtain a damage coefficient, and generating a low-risk signal and a high-risk signal according to the damage coefficient, thereby realizing intelligent graded early warning of the vehicle damage risk. According to the method, contact object type retrieval, scar feature matching and data statistics are carried out based on the integrated database, and data support is provided for analysis.

Description

Technical Field

[0001] This invention relates to the field of vehicle scratch detection technology, and in particular to a method and system for identifying and analyzing vehicle damage status based on multimodal data. Background Technology

[0002] With the rapid development of intelligent transportation systems, real-time identification and analysis of vehicle damage status is of great significance for vehicle safety maintenance, insurance loss assessment, and accident liability determination.

[0003] Prior art CN108090838A discloses a method, apparatus, server, client, and system for identifying damaged vehicle parts, including the pre-establishment of a feature library containing multiple identifiable vehicle parts and a feature correspondence library of the relative positional relationships of these vehicle parts. During the damage assessment image capture process, the user can manually circle the damaged location on the client. Then, the server can identify identifiable vehicle features in the image and determine its relative position to the user-circulated mark based on these identifiable features.

[0004] However, traditional vehicle damage detection methods mainly rely on manual inspection or single-modal data processing. For example, manually visually inspecting scratches, dents and other damage on the vehicle surface is inefficient, subjective, and difficult to detect hidden or minor damage. At the same time, there is insufficient correlation analysis between the type of contact object and the damage characteristics, making it difficult to accurately assess the damage risk. Consequently, there is a lack of systematic damage risk quantification methods, which cannot provide a scientific basis for vehicle maintenance and insurance decisions. Summary of the Invention

[0005] The purpose of this invention is to solve the problems in the background art by proposing a vehicle damage status identification and analysis method and system based on multimodal data.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A vehicle damage status identification and analysis method based on multimodal data, which specifically includes the following steps: Step 1: Monitor the target vehicle in real time, and generate a detection signal when a designated contact object is identified on the target vehicle; Step 2: When the detection signal is detected, video images between the specified contact object and the location where the anomaly occurs are acquired. According to the initial contact time, the preceding feature image and the termination image are acquired respectively. Based on the difference processing method, the difference image is obtained, and the anomaly signal is determined based on the difference image. Step 3: Identify the type of the specified contact object in the rear view, search and count in the integrated database to determine the number of damage occurrences and the total number of contacts of the corresponding type of contact object, and calculate to determine the probability of injury from the specified contact object; Step 4: Obtain the difference image and simultaneously retrieve the scratch features of the specified contact object from the integrated database. Then, compare the similarity between the difference image and the corresponding scratch features to determine the scratch similarity value. Step 5: Weight the probability of damage from the specified contact object with the similarity value of the scar to obtain the damage coefficient of the location where the anomaly occurred, and then generate low-risk and high-risk signals based on the damage coefficient.

[0007] As a further aspect of the present invention, the method for generating the detection signal includes: The monitoring equipment is installed on the target vehicle and used to monitor the target vehicle in real time. When the monitoring equipment detects that an object is in contact with the target vehicle, the corresponding object is marked as the designated contact object and a detection signal is generated at the same time. The target vehicle refers to the vehicle whose status needs to be monitored. The monitoring equipment includes image acquisition equipment and radar. The image acquisition equipment includes a front-view camera and a rear-view camera. The target vehicle is equipped with multiple monitoring devices, which are used to comprehensively cover and identify the appearance outline of the target vehicle.

[0008] As a further aspect of the present invention, the method for obtaining the preceding feature image and the terminating image includes: Identify the time of initial contact between the specified contact object and the target vehicle, and mark this time as the initial contact time; Identify the location where the specified contact object contacts the target vehicle and mark this location as the location of the anomaly. Set the initial contact time as the time node and collect the pre-contact and post-contact images of the location of the anomaly. The pre-contact image refers to the video image monitored within a fixed time t1 before the initial contact time, and the post-contact image refers to the video image collected in real time at the location of the anomaly starting from the initial contact time. Divide the fixed time t1 into units of time and set multiple time nodes. Identify the corresponding foreground image at each time node and label it as PBi, where i represents different time nodes and the unit of time is set to 5s. Using image processing tools, all the preceding images PBi are processed and integrated to obtain the preceding feature image of the anomaly location. Get the rear view, take the image corresponding to the final time in the rear view, and mark it as the termination image.

[0009] As a further aspect of the present invention, the end time of the acquisition of the rear view is the time when the specified contact object leaves the target vehicle. When the specified contact object is in a stationary state at the location where the abnormality occurs, the time when the stationary state occurs is identified. When the time when the stationary state occurs is greater than or equal to the time threshold A1, the time when the stationary state occurs is marked as the end time of the acquisition of the rear view.

[0010] As a further aspect of the present invention, the method for determining abnormal signals includes: Using image processing tools, the termination image and the preceding feature image are spatially aligned. Then, the termination image and the preceding feature image are subtracted pixel by pixel to obtain a difference image. When the difference image is 0, a normal image is generated, indicating that the location of the abnormality of the target vehicle has not been damaged. Conversely, when the difference image is not equal to 0, an abnormal signal is generated, and the state of the target vehicle at this time is marked as abnormal.

[0011] As a further aspect of the present invention, the method for determining the probability of injury from a specified contact object includes: Image recognition technology is used to identify the specified contact object in the rear view, determine the type of the specified contact object, search for the same type of contact object in the integrated database, and count the number of times the vehicle was damaged when the same type of contact object came into contact with the vehicle based on the search results, and mark it as the number of damage occurrences. Divide the number of times damage to the specified contact object occurs by the total number of times the corresponding contact object is contacted, and label the resulting value as the damage probability Gs of the specified contact object.

[0012] As a further aspect of the present invention, the integrated database stores experiments or instances of different types of contact objects coming into contact with vehicles, and the total number of contacts for each type of contact object is greater than or equal to the minimum sample size required for data calculation.

[0013] As a further aspect of the present invention, the method for determining the scar similarity value includes: The scratch features of the specified contact object are obtained from the integrated database, and the difference image is obtained at the same time. The similarity of the difference image and the scratch features of the specified contact object is compared to obtain the scratch similarity value Hs. The similarity comparison method adopts the ORB combined with template matching algorithm. By combining FAST key point detection and BRIEF descriptor, rotation invariance and pyramid structure are added, and then the normalized correlation coefficient is combined to calculate the pixel correlation between the difference image and the scratch features. The correlation value range is [-1,1].

[0014] As a further aspect of the present invention, the method for generating low-risk signals and high-risk signals includes: Obtain the damage probability Gs and the scar similarity value Hs of a specified contact object, and use the formula The damage coefficient Sx at the location of the anomaly is obtained, where, and These are the proportionality coefficients, and ; The damage coefficient Sx at the location of the anomaly is compared with the damage threshold Sy. If Sx < Sy, a low-risk signal is generated; otherwise, if Sx ≥ Sy, a high-risk signal is generated.

[0015] A vehicle damage status identification and analysis system based on multimodal data includes: An integrated database is used to store collision information and corresponding scratch features of various contact objects. The vehicle monitoring module is used to monitor the target vehicle in real time. When a specified contact object is identified on the target vehicle, a detection signal is generated. The anomaly recognition module is used to identify detection signals and acquire video images between the specified contact object and the location where the anomaly occurs. According to the initial contact time, the video images are divided into front and back frames. Based on the front and back frames, the front feature image and the termination image are acquired respectively. Based on the difference processing method, the difference image is obtained, and the anomaly signal is determined based on the difference image. The item analysis module is used to identify the type of a specified contact object in the rear view, and to search and count the data in the integrated database to determine the number of times damage occurs and the total number of contacts of the corresponding type of contact object, and to calculate and determine the probability of injury from the specified contact object. The image analysis module is used to acquire the difference image and retrieve the scratch features of the specified contact object from the integrated database. Then, the difference image is compared with the corresponding scratch features to determine the scratch similarity value. The comprehensive analysis module is used to weight the probability of injury from a specified contact object with the similarity value of the scar to obtain the damage coefficient at the location of the anomaly, and then generate low-risk and high-risk signals based on the damage coefficient; The terminal display module is used to output information such as the location of the anomaly, low-risk signals, and high-risk signals.

[0016] Compared with existing technologies, the advantages of this invention are: This invention monitors target vehicles in real time, immediately generating a detection signal upon identifying a designated contact object, and rapidly acquiring and analyzing video images. It then obtains a preceding feature image and a terminating image, performs interpolation processing to obtain a difference image, identifies the type of contact object, calculates the probability of injury by counting the number of damage occurrences and the total number of contacts, and compares the difference image with scar features to determine the scar similarity value. Finally, it obtains a damage coefficient by weighting the damage probability and scar similarity value, and generates low-risk and high-risk signals accordingly, achieving intelligent graded early warning of vehicle damage risk. This invention accurately determines the probability and severity of damage to vehicles caused by different contact objects through quantitative assessment. Furthermore, relying on an integrated database for contact object type retrieval, scar feature matching, and data statistics not only provides rich data support for analysis but also facilitates system updates and expansion with new contact object types and scar features, adapting to vehicle damage detection needs in different scenarios. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Reference Figure 1 and Figure 2 A vehicle damage status identification and analysis method based on multimodal data, which specifically includes the following steps: Step 1: Install monitoring equipment on the target vehicle and use the monitoring equipment to monitor the target vehicle in real time. When the monitoring equipment detects that an object is in contact with the target vehicle, it marks the corresponding object as the designated contact object and generates a detection signal at the same time. The target vehicle refers to the vehicle whose status needs to be monitored. The monitoring equipment includes image acquisition equipment and radar, etc. Furthermore, the image acquisition equipment includes front-view cameras and rear-view cameras, etc. In this embodiment, multiple monitoring devices are installed on the target vehicle, and the multiple monitoring devices are used to comprehensively cover and identify the appearance outline of the target vehicle. Step Two: When a detection signal is detected, identify the initial contact time between the designated contact object and the target vehicle, and mark this time as the initial contact time. Acquire video images of the contact positions before and after the initial contact time, and perform comparative analysis to determine the abnormal state of the target vehicle. Specific methods for determining the abnormal state include: S1: Identify the location where the specified contact object contacts the target vehicle, mark this location as the location of the anomaly, set the initial contact time as the time node, and collect the pre-contact and post-contact images of the location of the anomaly. The pre-contact image refers to the video image monitored within a fixed time t1 before the initial contact time, and the post-contact image refers to the video image collected in real time at the location of the anomaly starting from the initial contact time. Furthermore, the specific value of the fixed time t1 is set by those skilled in the art based on big data experience. In this embodiment, the fixed time t1 is set to 60s, that is, the video image of the location where the abnormality occurs within 60s before the initial contact time is the foreground image. Furthermore, the end time of the rear view acquisition is the time when the specified contact object leaves the target vehicle. In another embodiment of the present invention, when the specified contact object is in a stationary state at the location where the anomaly occurs, the time when the stationary state occurs is identified. When the time when the stationary state occurs is greater than or equal to the time threshold A1, the time when the stationary state occurs is marked as the end time of the rear view acquisition. The specific value of the time threshold A1 is set by those skilled in the art based on big data experience. The stationary state refers to the state in which the relative position between the specified contact object and the target vehicle does not change with time. S2: Divide the fixed time t1 into units of time and set multiple time nodes. Identify the corresponding front image under each time node and mark it as PBi, where i represents different time nodes. Furthermore, the specific value of the unit time is set by those skilled in the art based on big data experience. In this embodiment, the unit time is set to 5s. Using image processing tools, all the preceding images PBi are processed and integrated to obtain the preceding feature image of the anomaly location. In this embodiment, OpenCV is selected as the image processing tool, and the specific processing method is existing technology, so it will not be described in detail here. Get the rear view, take the image corresponding to the final time in the rear view, and mark it as the termination image; S3: Using image processing tools, the termination image and the preceding feature image are spatially aligned. Then, the termination image and the preceding feature image are subtracted pixel by pixel to obtain the difference image. When the difference image is 0, a normal image is generated, indicating that the location of the abnormality of the target vehicle has not been damaged. Conversely, when the difference image is not equal to 0, an abnormal signal is generated, and the state of the target vehicle at this time is marked as an abnormal state. Furthermore, the image processing tool used in this embodiment is OpenCV. The specific processing method is existing technology and will not be described in detail here. A difference image of 0 indicates that the termination image is exactly the same as the preceding feature image, and a difference image that is not equal to 0 indicates that there is a difference between the termination image and the preceding feature image. Step 3: When an abnormal signal is detected, the rear view is acquired, and the designated contact object in the rear view is further identified. Then, the designated contact object is searched and analyzed in the integrated database to determine the probability of injury from the designated contact object. The specific methods for determining the probability of injury include: Image recognition technology is used to identify the specified contact object in the rear view, determine the type of the specified contact object, search for the same type of contact object in the integrated database, and count the number of times the vehicle was damaged when the same type of contact object came into contact with the vehicle based on the search results, and mark it as the number of damage occurrences. Furthermore, the image recognition technology used in this embodiment is deep learning image recognition technology, such as the Transformer-based image recognition model. The specific recognition processing method is existing technology and will not be described in detail here. Then, divide the number of times damage to the specified contact object occurs by the total number of times the corresponding contact object is contacted, and mark the resulting value as the damage probability Gs of the specified contact object; It should be further noted that the integrated database stores experiments or instances of different types of contact objects coming into contact with vehicles, and the total number of contact times for each type of contact object is greater than or equal to the minimum sample size required for data calculation. The specific data for the minimum sample size is set by those skilled in the art based on big data experience. Step 4: Obtain the scratch features of the specified contact object from the integrated database, and simultaneously obtain the difference image. Compare the similarity between the difference image and the scratch features of the specified contact object to obtain the scratch similarity value Hs. The similarity comparison method uses the ORB combined with template matching algorithm. By combining FAST key point detection and BRIEF descriptor, rotation invariance and pyramid structure are added, and then combined with normalized correlation coefficient, the pixel correlation between the difference image and the scratch features is calculated. The correlation value ranges from [-1, 1], and the higher the correlation value, the higher the matching degree. It should be further explained that the integrated database stores the scratch features of various types of contact objects, and the scratch features of each type of contact object are processed by an artificial intelligence algorithm before being transmitted to the integrated database for data storage. Furthermore, the artificial intelligence algorithm selected in this embodiment is the convolutional neural network algorithm, and the feature processing of images by the convolutional neural network algorithm is an existing technology, which will not be elaborated here. Step 5: Obtain the damage probability Gs and the scar similarity value Hs of the specified contact object, and use the formula The damage coefficient Sx at the location of the anomaly is obtained, where, and These are the proportionality coefficients, and , and The specific value is obtained by those skilled in the art through big data calculation. Furthermore, when the damage coefficient Sx is larger, it indicates that the risk of damage to the abnormal location of the target vehicle is greater, and conversely, when the damage coefficient Sx is smaller, it indicates that the risk of damage to the abnormal location of the target vehicle is smaller. Then, the damage coefficient Sx at the location of the anomaly is compared with the damage threshold Sy. If Sx < Sy, a low-risk signal is generated; otherwise, if Sx ≥ Sy, a high-risk signal is generated. Finally, the location of the anomaly, along with the generated low-risk and high-risk signals, are simultaneously transmitted to the terminal display module.

[0020] A vehicle damage status identification and analysis system based on multimodal data includes: An integrated database is used to store collision information and corresponding scratch features of various contact objects, and to establish one-way communication connections with the object analysis module and the image analysis module respectively. The vehicle monitoring module is used to monitor the target vehicle in real time. When a specified contact object is identified on the target vehicle, a detection signal is generated and transmitted to the anomaly identification module. The anomaly recognition module is used to identify detection signals and acquire video images between the specified contact object and the location where the anomaly occurs. According to the initial contact time, the video images are divided into front and back images. Based on the front and back images, the front feature image and the termination image are acquired respectively. Based on the difference processing method, the difference image is obtained. At the same time, the anomaly signal is determined based on the difference image and transmitted to the image analysis module. Meanwhile, the back image is transmitted to the object analysis module. The item analysis module is used to identify the type of a specified contact object in the rear view, and to search and count in the integrated database to determine the number of times damage occurs and the total number of contacts of the corresponding type of contact object. It also calculates the probability of injury from the specified contact object and then transmits the probability of injury to the comprehensive analysis module. The image analysis module is used to acquire the difference image, and at the same time, retrieve the scratch features of the specified contact object from the integrated database. Then, the difference image is compared with the corresponding scratch features to determine the scratch similarity value, and then transmitted to the comprehensive analysis module. The comprehensive analysis module is used to weight the probability of injury from a specified contact object with the similarity value of the scar to obtain the damage coefficient of the abnormal location. Then, based on the damage coefficient, low-risk signals and high-risk signals are generated and transmitted to the terminal display module. The terminal display module is used to output information such as the location of the anomaly, low-risk signals, and high-risk signals.

[0021] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for identifying and analyzing vehicle damage status based on multimodal data, characterized in that, The method specifically includes the following steps: Step 1: Monitor the target vehicle in real time, and generate a detection signal when a designated contact object is identified on the target vehicle; Step 2: When the detection signal is detected, video images between the specified contact object and the location where the anomaly occurs are acquired. According to the initial contact time, the preceding feature image and the termination image are acquired respectively. Based on the difference processing method, the difference image is obtained, and the anomaly signal is determined based on the difference image. Step 3: Identify the type of the specified contact object in the rear view, search and count in the integrated database to determine the number of damage occurrences and the total number of contacts of the corresponding type of contact object, and calculate to determine the probability of injury from the specified contact object; Step 4: Obtain the difference image and simultaneously retrieve the scratch features of the specified contact object from the integrated database. Then, compare the similarity between the difference image and the corresponding scratch features to determine the scratch similarity value. Step 5: Weight the probability of injury from the specified contact object with the similarity value of the scar to obtain the damage coefficient of the location where the anomaly occurred, and then generate low-risk and high-risk signals based on the damage coefficient.

2. The vehicle damage state identification and analysis method based on multimodal data according to claim 1, characterized in that, Methods for generating detection signals include: The monitoring equipment is installed on the target vehicle and used to monitor the target vehicle in real time. When the monitoring equipment detects that an object is in contact with the target vehicle, the corresponding object is marked as the designated contact object and a detection signal is generated at the same time. The target vehicle refers to the vehicle whose status needs to be monitored. The monitoring equipment includes image acquisition equipment and radar. The image acquisition equipment includes a front-view camera and a rear-view camera. The target vehicle is equipped with multiple monitoring devices, which are used to comprehensively cover and identify the appearance outline of the target vehicle.

3. The vehicle damage state identification and analysis method based on multimodal data according to claim 1, characterized in that, Methods for obtaining the preceding feature image and the terminating image include: Identify the time of initial contact between the specified contact object and the target vehicle, and mark this time as the initial contact time; Identify the location where the specified contact object contacts the target vehicle and mark this location as the location of the anomaly. Set the initial contact time as the time node and collect the pre-contact and post-contact images of the location of the anomaly. The pre-contact image refers to the video image monitored within a fixed time t1 before the initial contact time, and the post-contact image refers to the video image collected in real time at the location of the anomaly starting from the initial contact time. Divide the fixed time t1 into units of time and set multiple time nodes. Identify the corresponding foreground image at each time node and label it as PBi, where i represents different time nodes and the unit of time is set to 5s. Using image processing tools, all the preceding images PBi are processed and integrated to obtain the preceding feature image of the anomaly location. Get the rear view, take the image corresponding to the final time in the rear view, and mark it as the termination image.

4. The vehicle damage state identification and analysis method based on multimodal data according to claim 3, characterized in that, The end time of the rear view acquisition is the time when the specified contact object leaves the target vehicle. When the specified contact object is stationary at the location where the anomaly occurs, the time when the stationary state occurs is identified. When the time when the stationary state occurs is greater than or equal to the time threshold A1, the time when the stationary state occurs is marked as the end time of the rear view acquisition.

5. The vehicle damage state identification and analysis method based on multimodal data according to claim 3, characterized in that, Methods for determining abnormal signals include: Using image processing tools, the termination image and the preceding feature image are spatially aligned. Then, the termination image and the preceding feature image are subtracted pixel by pixel to obtain a difference image. When the difference image is 0, a normal image is generated, indicating that the location of the abnormality of the target vehicle has not been damaged. Conversely, when the difference image is not equal to 0, an abnormal signal is generated, and the state of the target vehicle at this time is marked as abnormal.

6. The vehicle damage state identification and analysis method based on multimodal data according to claim 1, characterized in that, Methods for determining the probability of injury from a specified contact object include: Image recognition technology is used to identify the specified contact object in the rear view, determine the type of the specified contact object, search for the same type of contact object in the integrated database, and count the number of times the vehicle was damaged when the same type of contact object came into contact with the vehicle based on the search results, and mark it as the number of damage occurrences. Divide the number of times damage to the specified contact object occurs by the total number of times the corresponding contact object is contacted, and label the resulting value as the damage probability Gs of the specified contact object.

7. The vehicle damage state identification and analysis method based on multimodal data according to claim 6, characterized in that, The integrated database stores experiments or instances of different types of contact objects coming into contact with vehicles, and the total number of contacts for each type of contact object is greater than or equal to the minimum sample size required for data calculation.

8. The vehicle damage state identification and analysis method based on multimodal data according to claim 1, characterized in that, Methods for determining scar similarity values ​​include: The scratch features of the specified contact object are obtained from the integrated database, and the difference image is obtained at the same time. The similarity of the difference image and the scratch features of the specified contact object is compared to obtain the scratch similarity value Hs. The similarity comparison method adopts the ORB combined with template matching algorithm. By combining FAST key point detection and BRIEF descriptor, rotation invariance and pyramid structure are added, and then the normalized correlation coefficient is combined to calculate the pixel correlation between the difference image and the scratch features. The correlation value range is [-1,1].

9. The vehicle damage state identification and analysis method based on multimodal data according to claim 1, characterized in that, Methods for generating low-risk and high-risk signals include: Obtain the damage probability Gs and the scar similarity value Hs of a specified contact object, and use the formula The damage coefficient Sx at the location of the anomaly is obtained, where, and These are the proportionality coefficients, and ; The damage coefficient Sx at the location of the anomaly is compared with the damage threshold Sy. If Sx < Sy, a low-risk signal is generated; otherwise, if Sx ≥ Sy, a high-risk signal is generated.

10. A vehicle damage state identification and analysis system based on multimodal data, wherein the analysis system refers to the vehicle damage state identification and analysis method based on multimodal data as described in any one of claims 1-9, characterized in that, include: An integrated database is used to store collision information and corresponding scratch features of various contact objects. The vehicle monitoring module is used to monitor the target vehicle in real time. When a specified contact object is identified on the target vehicle, a detection signal is generated. The anomaly recognition module is used to identify detection signals and acquire video images between the specified contact object and the location where the anomaly occurs. According to the initial contact time, the video images are divided into front and back frames. Based on the front and back frames, the front feature image and the termination image are acquired respectively. Based on the difference processing method, the difference image is obtained, and the anomaly signal is determined based on the difference image. The item analysis module is used to identify the type of a specified contact object in the rear view, and to search and count the data in the integrated database to determine the number of times damage occurs and the total number of contacts of the corresponding type of contact object, and to calculate and determine the probability of injury from the specified contact object. The image analysis module is used to acquire the difference image and retrieve the scratch features of the specified contact object from the integrated database. Then, the difference image is compared with the corresponding scratch features to determine the scratch similarity value. The comprehensive analysis module is used to weight the probability of injury from a specified contact object with the similarity value of the scar to obtain the damage coefficient at the location of the anomaly, and then generate low-risk and high-risk signals based on the damage coefficient; The terminal display module is used to output information such as the location of the anomaly, low-risk signals, and high-risk signals.

Images