A Method and System for Detecting Road Sign Surface Deformation Based on Dynamic Video Sequences

Abstract

This invention discloses a method and system for detecting surface deformation of road signs based on dynamic video sequences. The method includes: acquiring road video data with timestamps and location information; automatically identifying traffic signs from the road video data and performing geometric anomaly detection; establishing a temporal tracking sequence for the road signs and confirming the presence of persistent anomalies through multi-frame analysis; if so, performing local texture consistency and local contour geometric analysis on the road signs to obtain fusion analysis results; performing a state assessment on the supporting equipment corresponding to the road signs to obtain state assessment results for the road sign supporting equipment; and determining the final surface deformation state and corresponding level of the road signs based on the fusion analysis results and the state assessment results of the road sign supporting equipment. By implementing the method of this invention, automatic detection and quantitative analysis of bending or deformation of road signs can be achieved without manual intervention, and timely output of early warning information can be provided.

Description

Technical Field

[0001] This invention relates to a road sign detection method, and more specifically to a road sign surface deformation detection method and system based on dynamic video sequences. Background Technology

[0002] In modern urban transportation systems, road signs serve as crucial infrastructure, providing drivers with directional guidance, road information, and safety reminders. The integrity and clarity of these signs directly impact vehicle safety and the effectiveness of traffic organization. However, in practice, signs are frequently damaged due to external environmental factors or human activities.

[0003] For example, in traffic accidents, out-of-control vehicles may directly collide with signs; collisions with large vehicles or construction vehicles may also exert significant external force on the sign's support rods and main body. This external force often causes signs to bend, twist, or even suffer partial structural damage. If these problems are not repaired and addressed promptly, deformed signs not only affect the integrity of information transmission but may also pose a series of potential traffic safety risks: bent signs may cause distortion of text, graphics, or symbols, making it difficult for drivers to accurately identify them at high speeds, resulting in incorrect navigation information; partial damage or deformation may lead to the loss of crucial information, affecting the driver's judgment; if the directional information is unclear or incorrect, especially in complex road conditions, drivers are prone to making incorrect operations, increasing the risk of secondary accidents such as rear-end collisions, sudden braking, or temporary lane changes; signs that are not promptly detected and repaired will have a long-term negative impact on traffic safety and order in that section of road, creating a management blind spot.

[0004] Currently, road sign inspections primarily rely on manual patrols, a method that is inefficient, covers a wide area, and is prone to overlooking problems. Furthermore, there is often a significant time delay between the discovery of damaged signs and their eventual repair, allowing safety hazards to persist for extended periods. Additionally, manual assessments lack standardized quantitative criteria, making results susceptible to subjective influences and failing to meet the demands of modern cities for refined and timely road traffic safety management.

[0005] Therefore, it is necessary to design a new method to automatically detect and quantitatively analyze the bending or deformation of road signs without human intervention, and to output early warning information in a timely manner. This would effectively overcome the limitations of existing inspection methods, ensure that road signs are always in optimal working condition, and safeguard urban traffic safety. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for detecting surface deformation of road signs based on dynamic video sequences.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a method for detecting surface deformation of road signs based on dynamic video sequences, comprising:

[0008] Acquire road video data with timestamps and location information;

[0009] Traffic signs are automatically identified from the road video data, and geometric anomalies are detected.

[0010] A time-series tracking sequence was established for the road signs, and multi-frame analysis was used to confirm whether there were any persistent anomalies in the road signs.

[0011] When the road sign exhibits persistent anomalies, local texture consistency and local contour geometry analysis are performed on the road sign to obtain the fusion analysis results;

[0012] A condition assessment is performed on the supporting equipment corresponding to the road sign to obtain the condition assessment results of the road sign supporting equipment;

[0013] Based on the fusion analysis results and the status assessment results of the road sign support equipment, the final surface deformation state and corresponding level of the road sign are determined.

[0014] The further technical solution is as follows: establishing a time-series tracking sequence for the road sign and confirming whether the road sign has persistent anomalies through multi-frame analysis includes:

[0015] A tracking identifier is created and maintained for each road sign by utilizing the position, appearance features, and spatial coherence of the preceding and following frames;

[0016] Based on the tracking identifier, the screening data of each frame is recorded in real time to form a time-series tracking sequence;

[0017] A sliding time window is used to analyze consecutive frames in the time-series tracking sequence to determine whether the road sign has a persistent anomaly.

[0018] The further technical solution is as follows: the appearance features include at least one of color distribution and morphological feature vector, and the screening data includes at least one of aspect ratio deviation, outer contour deformation degree, and texture anomaly index.

[0019] The further technical solution is as follows: the automatic identification of traffic signs from the road video data and the detection of geometric anomalies include:

[0020] Traffic sign frames that meet the requirements are identified and filtered from the road video data, and then size normalization and edge enhancement processing are performed to obtain road sign images;

[0021] The road sign image is compared with the geometric parameters of road signs in the standard model database to check the consistency of aspect ratio, edge straightness and outer contour shape in order to detect geometric anomalies.

[0022] The further technical solution is as follows: comparing the road sign image with the geometric parameters of road signs in the standard model database, and checking the consistency of aspect ratio, edge straightness, and outer contour shape to perform geometric anomaly detection, includes:

[0023] The road sign image is compared with the geometric parameters of road signs in the standard model database. The length and width ratio of the detected object are compared to see if they meet the set ratio requirements. The straightness of the edge is evaluated by fitting a straight line equation through the boundary detection results. The consistency of the shape is verified by extracting and matching the outer contour.

[0024] The further technical solution is as follows: The local texture consistency and local contour geometry analysis of the road sign are performed to obtain the fusion analysis result, including:

[0025] Analyze the local texture of the road sign surface to determine if there are areas of abnormal stretching, twisting or breakage. If the local texture direction does not meet the set requirements, it is determined that the road sign has potential bending signs, so as to obtain the local texture consistency analysis results.

[0026] The overall outline of the road sign is extracted from the continuous frame sequence of the time-tracking sequence, and the shape changes under different viewpoints are compared. When there is a prominent deviation that does not conform to normal perspective transformation, it is determined that the road sign has structural deformation, so as to obtain the outline geometric analysis results.

[0027] The comprehensive evaluation result is calculated based on the local texture consistency analysis result and the contour geometry analysis result to obtain the fusion analysis result.

[0028] The further technical solution is as follows: The condition assessment of the supporting equipment corresponding to the road sign, to obtain the condition assessment result of the road sign supporting equipment, includes:

[0029] Based on the location of the road sign, the corresponding supporting equipment is extracted;

[0030] The main axis of the support device is calculated using a straight line detection algorithm, and the main axis of the support device is compared with a preset vertical direction to obtain a comparison result;

[0031] If the comparison result shows that the support device is tilted, bent, or broken, the support device is marked as abnormal to obtain a status assessment result for the road sign support device. If the comparison result does not show that the support device is tilted, bent, or broken, the support device is marked as normal to obtain a status assessment result for the road sign support device.

[0032] The further technical solution is as follows: after calculating the main axis of the support device using a straight line detection algorithm and comparing the main axis of the support device with a preset vertical direction to obtain the comparison result, it further includes:

[0033] When a portion of the support device is obscured or cannot be clearly identified due to insufficient image resolution, the reliability and confidence of the unidentifiable portion of the support device are reduced.

[0034] The further technical solution is as follows: Based on the fusion analysis results and the condition assessment results of the road sign support equipment, the final surface deformation state and corresponding level of the road sign are determined, including:

[0035] When the road sign is in an abnormal state, the final surface deformation state of the road sign is determined based on the fusion analysis results and the state assessment results of the road sign support equipment.

[0036] The grade is determined based on the final surface deformation state of the road sign.

[0037] This invention also provides a road sign surface deformation detection system based on dynamic video sequences, comprising:

[0038] The acquisition unit is used to acquire road video data with timestamps and location information;

[0039] The initial screening unit is used to automatically identify traffic signs from the road video data and perform geometric anomaly detection;

[0040] A consistency determination unit is used to establish a time-series tracking sequence for the road sign and to confirm whether the road sign has a persistent anomaly through multi-frame analysis;

[0041] The fusion analysis unit is used to perform local texture consistency and local contour geometry analysis on the road sign to obtain the fusion analysis result when the road sign has a persistent anomaly.

[0042] A status assessment unit is used to assess the status of the supporting equipment corresponding to the road sign in order to obtain the status assessment result of the road sign supporting equipment.

[0043] The comprehensive judgment unit is used to determine the final surface deformation state and corresponding level of the road sign based on the fusion analysis results and the state assessment results of the road sign support equipment.

[0044] The advantages of this invention compared to existing technologies are as follows: This invention acquires road video data with timestamps and location information, automatically identifies traffic signs and detects their geometric anomalies, establishes a time-series tracking sequence for the road signs, and confirms the existence of persistent anomalies through multi-frame analysis. For road signs with persistent anomalies, local texture consistency and contour geometry analysis are performed to obtain fusion analysis results. Simultaneously, the status of the supporting equipment is evaluated, and finally, based on these analyses, the surface deformation state and level of the road signs are determined. This process requires no manual intervention, achieving automatic detection and quantitative analysis of road sign bending or deformation, and timely output of early warning information. It effectively overcomes the limitations of traditional inspection methods, which are time-consuming, labor-intensive, and difficult to monitor in real time, ensuring that road signs are always in optimal working condition, thereby guaranteeing urban traffic safety.

[0045] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 A flowchart illustrating the method for detecting surface deformation of road signs based on dynamic video sequences provided in an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of a timing tracing sequence provided in an embodiment of the present invention. Figure 1 ;

[0049] Figure 3 A schematic diagram of a timing tracing sequence provided in an embodiment of the present invention. Figure 2 ;

[0050] Figure 4 A schematic diagram illustrating road sign recognition and comparison provided in an embodiment of the present invention;

[0051] Figure 5 A schematic diagram of the support device provided in an embodiment of the present invention;

[0052] Figure 6 A schematic block diagram of a road sign surface deformation detection system based on dynamic video sequences provided in an embodiment of the present invention;

[0053] Figure 7 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0056] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0057] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0058] Please see Figure 1 , Figure 1 This is a flowchart illustrating the method for detecting road sign surface deformation based on dynamic video sequences provided in this invention. The method is implemented in a server. The server interacts with terminals and cameras, acquiring road video data with timestamps and location information to automatically identify traffic signs and perform geometric anomaly detection. Once a road sign is detected, the system establishes a time-series tracking sequence and confirms persistent anomalies through multi-frame analysis. For road signs with persistent anomalies, further local texture consistency and local contour geometric analysis are performed to obtain fusion analysis results, and the status of the road sign support equipment is evaluated. Based on the above analysis results, the final surface deformation state and corresponding level of the road sign are determined. The entire process utilizes dynamic video sequences to achieve real-time monitoring and early warning of road sign bending or deformation, effectively solving the problems of low efficiency and narrow coverage in traditional inspection methods, ensuring that road signs are always in optimal working condition, thereby guaranteeing urban traffic safety. This method significantly improves the accuracy and timeliness of monitoring, and reduces maintenance costs and manpower requirements.

[0059] Figure 1This is a flowchart illustrating the method for detecting surface deformation of road signs based on dynamic video sequences provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following steps S110 to S160.

[0060] S110. Obtain road video data with timestamps and location information.

[0061] In this embodiment, road video data refers to a continuous sequence of images of the road and its surrounding environment captured in real time by a visual acquisition module installed on a vehicle or other mobile or fixed equipment. This video data not only includes specific images of traffic signs on both sides and above the road, but also includes precise timestamps and geographic location information to facilitate subsequent processing and analysis.

[0062] Specifically, acquiring road video data with timestamps and location information includes the following aspects:

[0063] Using high-definition cameras and other visual acquisition devices, videos are recorded at a certain frame rate (e.g., 30 frames per second) to ensure that road sign details are clearly visible. Each video segment is accompanied by GPS coordinates to determine its geographical location, and each frame is timestamped with precision down to the second or even finer granularity via a built-in clock.

[0064] The acquired video data is transmitted wirelessly to a central server for storage and initial processing. This step may involve video format conversion, compression, and metadata processing to ensure that the data can be efficiently used by subsequent algorithms.

[0065] Timestamps are used to synchronize data streams from different sources, enabling the system to react to specific events at the correct time. For example, in multi-frame consistency analysis, accurate time information helps identify and track changes in road sign status over time. Geographic location information helps pinpoint detection results to specific road segments or areas, providing important guidance for subsequent maintenance work.

[0066] Considering the various challenges that may be encountered in practical applications, such as changes in lighting and weather conditions, the acquired video data needs to have sufficient resolution and dynamic range to effectively capture the state of road signs in various environments. Furthermore, the system needs to design corresponding algorithms to overcome these interference factors and ensure detection accuracy.

[0067] In summary, acquiring road video data with timestamps and location information is fundamental to the entire technical process. It provides the necessary input for subsequent road sign recognition, rapid initial screening, fine-grained morphological fusion analysis, and support pole structural integrity detection. This process not only requires efficient video acquisition capabilities but also needs to ensure data integrity and accuracy to support the automated and intelligent operation of the system.

[0068] S120. Automatically identify traffic signs from the road video data and perform geometric anomaly detection.

[0069] In this embodiment, traffic signs refer to subsequent road signs.

[0070] In one embodiment, step S120 described above may include steps S121 to S122.

[0071] S121. Identify and filter traffic sign frames that meet the requirements from the road video data, and perform size standardization and edge enhancement processing to obtain road sign images.

[0072] In this embodiment, road sign images refer to one or more frames that have undergone preliminary screening and preprocessing, containing clearly visible road signs. These images not only remove interference factors such as occlusion and uneven lighting, but also undergo necessary preprocessing (such as size normalization and edge enhancement) to facilitate accurate comparison with standard models.

[0073] First, a deep learning object detection network is used to identify traffic signs in each frame of a continuous video stream captured by a vehicle camera. The model output includes the bounding box location, category, and confidence score for each detected road sign.

[0074] The filtering criteria include:

[0075] Unobstructed and clear: Ensure that the road sign area is not partially obscured by other objects, guaranteeing sufficient image quality for accurate analysis.

[0076] Uniform brightness and normal contrast: Avoid image distortion caused by overexposure or low light environments.

[0077] Pixel density is higher than a preset threshold: This ensures sufficient detail for geometric feature analysis.

[0078] The selected road sign images will be resized to a uniform size and their contour features will be enhanced using an edge detection algorithm, thus facilitating a consistent comparison with templates in the standard model database.

[0079] S122. Compare the road sign image with the geometric parameters of road signs in the standard model database to check the consistency of aspect ratio, edge straightness and outer contour shape, so as to perform geometric anomaly detection.

[0080] In this embodiment, the road sign image is compared with the geometric parameters of road signs in the standard model database to compare whether the aspect ratio of the detected object meets the set ratio requirements. The straightness of the edge is evaluated by fitting a straight line equation through the boundary detection results, and the consistency of the shape is verified by extracting and matching the outer contour.

[0081] This step aims to detect whether the road sign is deformed or has other anomalies by comparing detailed geometric parameters.

[0082] Aspect Ratio Verification: Calculate the width-to-height ratio of each road sign image and compare it to the ideal ratio of the corresponding type of road sign in the standard model database. If the deviation exceeds a set threshold (e.g., ±8%), it is marked as an aspect ratio anomaly.

[0083] Edge straightness analysis: Based on the boundary detection results, the straight line equations of the upper and lower edges of the road sign are fitted. If the residual or curvature of the fitted straight line exceeds a set threshold, it indicates that there is a risk of bending or local deformation in the shape.

[0084] Outer contour alignment: Extract the overall outer contour of the road sign and match it with the ideal contour shape in the standard model database. If the overlap between the two is significantly lower than the standard value, the road sign is considered to have a possible geometric anomaly.

[0085] Rapid anomaly detection and data labeling: Based on the results of the geometric analysis above, the system generates rapid identification labels (normal, suspected anomaly, or insufficient data) for each road sign frame. These label information will be recorded and passed as input to the subsequent multi-frame tracking and consistency determination module for cumulative analysis over time.

[0086] By accurately identifying and filtering road sign images from complex road video data, and then performing detailed geometric parameter comparisons, geometric anomaly detection of road signs can be effectively achieved. This lays the foundation for further refined morphological fusion analysis and structural integrity assessment of support rods. This process not only improves the accuracy of detection but also enhances the system's robustness and adaptability to complex environments.

[0087] S130. Establish a time-series tracking sequence for the road sign, and confirm whether there is a persistent anomaly in the road sign through multi-frame analysis.

[0088] In one embodiment, step S130 described above may include steps S131 to S133.

[0089] S131. Create and maintain a tracking identifier for each road sign by utilizing the position, appearance features, and spatial coherence of the preceding and following frames.

[0090] In this embodiment, appearance features include, but are not limited to, color distribution and morphological feature vectors.

[0091] Record the precise location coordinates (such as the center point coordinates of the bounding box) of the road sign target detected in each frame to determine its motion trajectory in consecutive frames.

[0092] Specifically, the appearance features include at least one of color distribution and morphological feature vector.

[0093] Color distribution: Extract color histogram or color space model (such as HSV) features from road signs to identify and track the same target.

[0094] Morphological feature vector: Calculate the shape features of the road sign (such as contour descriptor, Hu moment, etc.) to ensure consistent tracking from different viewpoints.

[0095] Spatial coherence: Based on the spatial position change patterns of the target (such as velocity, acceleration, etc.), the position of the target in the current frame is predicted by combining information from previous and subsequent frames, thereby improving the accuracy of tracking.

[0096] Using the above method, the system can create a unique tracking identifier for each road sign and maintain the identifier to achieve target association across frames.

[0097] S132. Based on the tracking identifier, record the screening data of each frame in real time to form a time-series tracking sequence.

[0098] In this embodiment, the screening data includes at least one of aspect ratio deviation, outer contour deformation, and texture anomaly index.

[0099] like Figures 2 to 3 As shown, the temporal tracking sequence refers to the system continuously tracking road signs based on the detection results of previous and subsequent frames, through target position, appearance features and spatial coherence, thereby establishing a trajectory record of each road sign changing over time in the video stream, which is used to support multi-frame consistency analysis and anomaly detection.

[0100] In this embodiment, for each tracked road sign, the relevant screening data is updated in each frame based on its latest status. For example:

[0101] Aspect Ratio Deviation: Calculates the difference in width and height ratio between the current frame and the standard model.

[0102] Outer contour deformation: The straightness of the edge is evaluated by fitting a straight line equation to the boundary detection results and comparing it with the ideal contour shape.

[0103] Texture anomaly index: Analyze the consistency of the surface texture of road signs to find possible abnormal stretching, distortion or breakage areas.

[0104] A time-series tracking sequence refers to the set of all relevant screening data arranged in chronological order for each road sign target. This sequence not only includes the specific measurements of each frame, but also reflects the trends and fluctuations over time, providing fundamental data support for subsequent multi-frame consistency analysis.

[0105] S133. Use a sliding time window to analyze consecutive frames in the time-series tracking sequence to determine whether the road sign has a persistent anomaly.

[0106] Select a fixed-length time window (e.g., N frames), and slide it step by step along the time-series tracing sequence, performing consistency analysis on consecutive frames within each window. The specific steps are as follows:

[0107] Within a selected time window, count the number of all frames containing anomaly markers and their specific anomaly types (such as aspect ratio anomalies, outer contour deformation, etc.).

[0108] If the same type of anomaly appears in most frames (e.g., more than 3 out of 5 frames), the anomaly is considered persistent; otherwise, if the anomaly is only displayed in a few frames, it is considered transient noise and ignored.

[0109] This multi-frame consistency determination method based on time accumulation can effectively distinguish between genuine deformation or damage and transient interference caused by environmental factors, thereby improving the robustness and reliability of the system. Finally, the system will trigger a refined analysis module based on this result to further verify the true state of suspected abnormal targets.

[0110] In this embodiment, geometric anomaly detection is performed first, followed by persistent anomaly detection. Only road signs that meet the criteria of both anomalies need to undergo subsequent fusion analysis.

[0111] S140. When the road sign has a persistent anomaly, perform local texture consistency and local contour geometry analysis on the road sign to obtain the fusion analysis result.

[0112] In this embodiment, the fusion analysis result refers to the comprehensive evaluation of the results of two methods—local texture consistency analysis and contour geometry analysis—on the road sign surface to determine whether the road sign has actual deformation or damage. This result not only improves the accuracy of single-method judgments but also reduces the possibility of misjudgment through cross-validation of data from different dimensions, providing more reliable detection conclusions.

[0113] In one embodiment, step S140 described above may include steps S141 to S143.

[0114] S141. Analyze the local texture of the road sign surface to determine whether there are areas of abnormal stretching, twisting or breakage. If the local texture direction does not meet the set requirements, it is determined that the road sign has potential bending signs, so as to obtain the local texture consistency analysis results.

[0115] In this embodiment, the local texture consistency analysis result refers to the system's detailed analysis of the local texture of the road sign surface, identifying potential anomalies by comparing its consistency with the overall texture trend. For example, if a portion of the texture shows abnormal stretching, distortion, or breakage, and these changes significantly deviate from the normal texture trend, then this area is considered to potentially have physical deformation or damage. This analysis helps identify subtle deformations or early damage on the road sign that are not easily noticeable.

[0116] S142. Extract the overall outline of the road sign from the continuous frame sequence of the time-tracking sequence, and compare the shape changes under different viewpoints. When a region with a prominent deviation that does not conform to normal perspective transformation appears, it is determined that the road sign has structural deformation, so as to obtain the outline geometric analysis result.

[0117] In this embodiment, the contour geometry analysis result refers to the system comparing the contours from different viewpoints based on the road sign contour information extracted from a multi-frame video sequence, and searching for change points that do not follow normal perspective transformation rules. These change points may indicate that certain parts of the road sign have undergone structural deformation. For example, if the contour of a certain area shows unusual curvature or protrusion in multiple viewpoints, this usually means that there may be serious physical deformation in that area, such as bending or depression.

[0118] S143. Calculate the comprehensive evaluation result based on the local texture consistency analysis result and the contour geometry analysis result to obtain the fusion analysis result.

[0119] In this step, the system combines the local texture consistency analysis results obtained in S141 with the contour geometry analysis results obtained in S142, and calculates a comprehensive evaluation result using a certain algorithm (such as confidence-weighted fusion). This fusion analysis result aims to leverage the complementary advantages of the two analysis methods to provide a more comprehensive and accurate road sign condition assessment. This not only allows for more precise location of the road sign's deformation and degree but also provides a scientific basis for subsequent maintenance decisions.

[0120] like Figure 4 As shown, the system first acquires a road sign image and compares it with a standard road sign model stored in the database. The metrics used in the matching process include, but are not limited to, aspect ratio, edge straightness, and outer contour shape. If the acquired road sign image deviates significantly from the standard values ​​in width and height ratios, or if its outer contour shows bending or distortion, it indicates a significant difference between the road sign and the normal model, and the road sign is marked as "suspected bent." At this point, the system saves the data for further analysis, ensuring that this key data can be used when determining the consistency of multiple frames.

[0121] In this embodiment, step S140 performs a more in-depth morphological analysis on these suspected abnormal targets to improve the accuracy and robustness of judging road sign bending or deformation. This process mainly includes local texture consistency analysis, contour geometry analysis, and confidence-weighted fusion of the results.

[0122] First, a detailed texture analysis was performed on each suspected abnormal road sign. This involved comparing texture patterns in different areas of the road sign surface, paying particular attention to any abnormal stretching, twisting, or breakage. These phenomena could be caused by physical damage, such as impact or material aging due to prolonged exposure to harsh environments.

[0123] If a local texture direction deviates significantly from the overall texture trend, it could be a signal that there is potential bending or deformation in that area. This analytical method can capture minute changes that are difficult to detect with the naked eye, thus providing early warning.

[0124] For each road sign flagged as potentially abnormal, its overall outline is extracted from its consecutive frame sequences. The system then compares the outline changes from different perspectives (e.g., shots taken from different angles). This process considers not only changes in the front view of the road sign but also views from the side or other angles to gain a comprehensive understanding of the road sign's condition.

[0125] If a significant deviation is found in a local area during contour comparison from multiple perspectives, and this deviation cannot be explained by normal perspective transformation, it can be concluded that structural deformation has occurred in that area. This method is helpful in identifying more serious physical damage, such as bending or denting.

[0126] To further improve detection accuracy, the system integrates the results of the two analysis methods mentioned above—local texture consistency analysis and contour geometry analysis. Each method is assigned a corresponding weight based on its own reliability and confidence level. For example, if texture analysis provides more explicit information in a certain situation, it is given a higher weight; conversely, if it does not, it is given a lower weight.

[0127] Based on the confidence levels of each method, the system calculates a final comprehensive judgment result. This aims to avoid the risk of misjudgment that might arise from relying on a single method, ensuring that the final conclusion is as close as possible to the true situation. Furthermore, this method enhances the system's robustness, enabling it to maintain high detection accuracy even in complex environments.

[0128] Through the meticulous analysis process and result fusion strategy described above, the system can more reliably identify the actual curvature of road signs, providing high-confidence data support for subsequent maintenance decisions. This method not only improves detection efficiency but also significantly reduces the false alarm rate, making intelligent monitoring of road signs possible.

[0129] S150. Perform a status assessment on the supporting equipment corresponding to the road sign to obtain the status assessment result of the road sign supporting equipment.

[0130] In this embodiment, the status assessment result of the road sign support device refers to the determination of whether the support device is in a normal or abnormal state based on the deviation and integrity analysis of the main axial direction of the support rod from the preset vertical direction.

[0131] like Figure 5 As shown, the system automatically extracts the area where the support rod is located from the image and draws a baseline as a reference standard. Next, the system calculates the angle between the main axis of the support rod and this baseline to assess whether the support rod exhibits significant tilting or bending. If the calculated angle exceeds a preset threshold, it indicates a structural abnormality in the support rod, and the system marks the support rod as an "abnormal structure" in the results. This method effectively identifies potential problems with the support rod, providing accurate information support for subsequent maintenance work. This automated detection method not only improves detection efficiency but also ensures a comprehensive assessment of the safety of road signs and their supporting structures.

[0132] In one embodiment, step S150 described above may include steps S151 to S153.

[0133] S151. Based on the location of the road sign, extract the supporting equipment corresponding to the road sign.

[0134] First, the system needs to extract the corresponding supporting equipment (such as support poles) based on the identified and located road sign positions. This step mainly relies on image processing techniques, including but not limited to image segmentation and edge detection methods, to accurately separate the road sign and the support pole area below or to the side from the video frame.

[0135] By analyzing differences in color, texture, or brightness within the image, the road sign is distinguished from the background and the support pole area. Edge detection methods are then used to accurately delineate the outlines of the road sign and support pole, providing clear geometric information for subsequent analysis.

[0136] S152. Calculate the main axis of the support device using a straight line detection algorithm, and compare the main axis of the support device with a preset vertical direction to obtain a comparison result.

[0137] In this embodiment, the comparison result refers to determining whether the support rod is tilted, bent, or broken by calculating the angular difference between the main axis of the support device and the ideal vertical direction.

[0138] Next, the system uses a line detection algorithm (such as the Hough transform) to calculate the main axial direction of the support structure. This step aims to quantify the angular deviation of the support rod relative to the ideal vertical direction, thereby determining whether it has tilted, bent, or broken.

[0139] The main axis of the support rod is identified using a straight line detection method.

[0140] Calculate the angle between the axis and the vertical baseline. If the angle exceeds a set threshold, a potential problem is considered.

[0141] The calculated angle is compared with the preset standard. If the angle deviation exceeds a certain range (e.g., ±3 degrees), it is preliminarily determined that there is tilting or bending; if obvious discontinuities or breaks are found, it may indicate that the support rod is at risk of breaking.

[0142] S153. If the comparison result shows that the supporting equipment is tilted, bent, or broken, the supporting equipment is marked as abnormal to obtain a status assessment result of the road sign supporting equipment. If the comparison result does not show that the supporting equipment is tilted, bent, or broken, the supporting equipment is marked as normal to obtain a status assessment result of the road sign supporting equipment.

[0143] Finally, based on the comparison results in S152, the system will mark the supporting equipment as normal or abnormal.

[0144] When the main axis of the support equipment shows a significant deviation (i.e., greater than the set threshold) or signs of breakage are detected, the system marks it as "abnormal," which means that there may be a safety hazard and further inspection or repair is required.

[0145] If the comparison results show that the support rod remains in a near-ideal vertical direction and no evidence of any physical damage is found, it is marked as "normal".

[0146] In one embodiment, step S150 above further includes:

[0147] When a portion of the support device is obscured or cannot be clearly identified due to insufficient image resolution, the reliability and confidence of the unidentifiable portion of the support device are reduced.

[0148] In addition, to improve the accuracy of the assessment, the system can also combine other factors for a comprehensive evaluation, such as the visibility and integrity of the support rods. When part of the support rod is obscured or the resolution is insufficient to affect the judgment, the system may lower the confidence level but still record relevant information as an auxiliary reference.

[0149] Specifically, when the support rod is partially obscured or has insufficient resolution, the system will identify this situation and accordingly lower its confidence level in assessing the support rod's condition. This means that although the system can still attempt to detect and record certain features (such as the support rod's tilt angle and degree of bending), the accuracy of these features may be affected due to the quality of the input data. Therefore, the system marks this information as low confidence, prompting the user to treat this information with caution when making a final judgment.

[0150] Despite the aforementioned limitations, the system still records all identifiable features and provides them to the user as supplementary reference information. This is done to reflect the actual condition of the support rod as comprehensively as possible, providing useful information even under less than ideal conditions. For example, even if the image quality is insufficient to determine the exact tilt angle of the support rod, the system can still provide a general directional conclusion by analyzing the visible portions, aiding in further diagnosis.

[0151] By directly inspecting the support rod structure, the system can provide additional dimensions for judgment beyond the analysis of the card's shape. This includes, but is not limited to, assessments of the support rod's straightness, verticality, and overall structural integrity. This multi-dimensional analysis method not only increases the amount of diagnostic information but also improves the reliability of the overall diagnosis.

[0152] like Figure 5 As shown, by detecting the support rod structure, the system can provide an additional dimension of judgment beyond the analysis of the card face shape, further improving the reliability of the overall diagnosis.

[0153] S160. Based on the fusion analysis results and the status assessment results of the road sign support equipment, determine the final surface deformation state and corresponding level of the road sign.

[0154] In one embodiment, step S160 described above may include steps S161 to S162.

[0155] S161. When the road sign is in an abnormal state, the final surface deformation state of the road sign is determined based on the fusion analysis results and the state assessment results of the road sign support equipment.

[0156] In this embodiment, the final surface deformation state of the road sign refers to any abnormalities in the overall shape of the road sign obtained through comprehensive analysis. Specifically:

[0157] Local or global curvature: The presence of local or global curvature on road signs is identified through fine-grained morphological fusion analysis (including contour geometry analysis and local texture consistency analysis).

[0158] Local deformation: Identify whether there are obvious stretching, twisting or breakage in local areas on the road sign.

[0159] Abnormal support rod structure: By combining the analysis of the straightness, inclination and visibility of the support rod, it is determined whether the support rod is bent, broken or tilted, and this is used as auxiliary reference information.

[0160] These analytical results will be integrated to form a comprehensive diagnostic conclusion that considers not only the morphological characteristics of the sign itself, but also the integrity of the supporting structure, thereby more accurately reflecting the actual condition of the road sign.

[0161] S162. Determine the grade based on the final surface deformation state of the road sign.

[0162] In this embodiment, the level refers to the classification of the severity of the road sign status based on the above comprehensive analysis. The specific level division can be as follows:

[0163] Normal: No abnormalities were found in the road sign and support pole, indicating that they are in good working condition.

[0164] Anomalies (low confidence): Some detection metrics show certain deviations, but the overall information is insufficient to make a high-confidence anomaly judgment, or anomalies are only observed in a few frames. In this case, further monitoring is required.

[0165] Anomalies (high confidence): Multiple analyses consistently show significant bending or deformation, and these anomalies persist across multiple frame sequences, indicating a clear problem with the road sign that requires timely attention.

[0166] Unable to assess: The system is unable to provide reliable assessment results due to insufficient data clarity or other reasons. This usually occurs when the video quality is poor or the target is severely occluded.

[0167] Furthermore, based on the final diagnostic results and confidence levels, the system will also generate a corresponding risk level (e.g., low, medium, high) for each road sign, enabling traffic management departments to make appropriate repair or replacement decisions based on this quantitative data. This process ensures the accuracy, reliability, and traceability of road sign status assessments, while also improving the efficiency and intelligence of road traffic safety management.

[0168] In this embodiment, the system evaluates the reliability (i.e., confidence level) of the above-mentioned results and assigns weights based on their importance and accuracy. For example, if multi-frame consistency analysis shows a high-confidence anomaly, this result will have a larger weight in the overall score; conversely, if a single indicator deviates but the overall information is insufficient, its weight will be lower. This ensures that the overall score can comprehensively and accurately reflect the true state of the road signs.

[0169] Based on the result category and confidence level, the system generates a corresponding risk level (low, medium, high) for subsequent management or maintenance scheduling.

[0170] The risk level is determined based on a comprehensive score and conclusion classification. For example, a road sign in a "normal" state has a risk level of "low"; for a road sign that is "abnormal (high confidence)," considering that it may pose a greater safety hazard, the system may assign a higher risk level (such as "high"); while a road sign that is "abnormal (low confidence)" may be rated as a "medium" risk level.

[0171] In this way, the system can not only generate current road sign status reports but also save historical data, forming a detailed and traceable record system. This is of great significance for long-term tracking of road sign status changes, optimizing maintenance plans, and improving road traffic safety management. Furthermore, this tiered diagnostic result helps traffic management departments rationally allocate resources based on actual conditions, achieving efficient and accurate monitoring and maintenance of road signs.

[0172] The method in this embodiment addresses the challenges of automatic detection of bending deformation, robust analysis and anomaly suppression across multiple frame sequences, and fusion judgment of multi-source results—problems that cannot be achieved with existing technologies. First, through rapid geometric screening, multi-frame consistency analysis, and refined morphological fusion (including contour geometry and texture consistency), automatic detection and severity assessment of road sign deformation are achieved, maintaining high accuracy even under side-mounted, obliquely mounted, or partially occluded conditions. Second, a multi-frame consistency judgment mechanism is established using road sign tracking sequences and time-cumulative analysis, effectively distinguishing between transient anomalies and true deformation, thus improving the stability and robustness of the detection process. Furthermore, this technology integrates morphological features, texture features, and the structural state of the support rod for comprehensive analysis, and employs a confidence-weighted fusion strategy to ensure the objectivity and reliability of the final judgment result.

[0173] Specifically, by combining rapid road sign screening, multi-frame consistency determination, and fine-grained morphological fusion analysis, this technology enables the quantitative assessment of local and overall bending or deformation of road signs. This method utilizes texture consistency and contour geometry analysis to ensure the accuracy and reliability of detection even when road signs are side-mounted, angled, or partially obscured.

[0174] When processing video sequences, temporal cumulative analysis of consecutive frame sequences and multi-frame consistency determination methods are employed. This enables the system to effectively distinguish between transient noise and actual anomalies, thereby reducing the false positive rate. When encountering situations where data clarity is insufficient or anomaly information in consecutive frames is inadequate, the system outputs a low-confidence result or a "cannot be evaluated" conclusion to ensure its robustness in complex environments.

[0175] Furthermore, it integrates sign morphology analysis and support rod structural integrity testing to provide hierarchical diagnostic conclusions, along with confidence scores and risk levels, thereby achieving a comprehensive and multi-dimensional assessment of road sign status. This method provides quantitative decision support for traffic management and is applicable to real-world application scenarios involving video collection across multiple road segments and various vehicle types.

[0176] Compared to traditional manual inspections and single detection methods, this system significantly improves the efficiency of automated analysis of road sign deformation, structural anomalies, and risk levels, while reducing reliance on human labor. Therefore, it is highly suitable for large-scale deployment in road traffic management and intelligent monitoring systems.

[0177] The aforementioned method for detecting road sign surface deformation based on dynamic video sequences automatically identifies traffic signs and detects their geometric anomalies by acquiring road video data with timestamps and location information. It establishes a temporal tracking sequence for the road signs and uses multi-frame analysis to confirm the existence of persistent anomalies. For road signs with persistent anomalies, it performs local texture consistency and contour geometry analysis to obtain fusion analysis results. Simultaneously, it assesses the status of supporting equipment and finally determines the surface deformation state and level of the road signs based on these analyses. This process requires no manual intervention, achieving automatic detection and quantitative analysis of road sign bending or deformation, and can promptly output early warning information. It effectively overcomes the limitations of traditional inspection methods, which are time-consuming, labor-intensive, and difficult to monitor in real time, ensuring that road signs are always in optimal working condition, thereby guaranteeing urban traffic safety.

[0178] Figure 6 This is a schematic block diagram of a road sign surface deformation detection system 300 based on dynamic video sequences provided in an embodiment of the present invention. Figure 6 As shown, corresponding to the above-described method for detecting road sign surface deformation based on dynamic video sequences, the present invention also provides a road sign surface deformation detection system 300 based on dynamic video sequences. This road sign surface deformation detection system 300 based on dynamic video sequences includes a unit for executing the above-described method for detecting road sign surface deformation based on dynamic video sequences, and the system can be configured in a server. Specifically, please refer to... Figure 6 The road sign surface deformation detection system 300 based on dynamic video sequences includes an acquisition unit 301, a preliminary screening unit 302, a consistency determination unit 303, a fusion analysis unit 304, a status evaluation unit 305, and a comprehensive determination unit 306.

[0179] The system comprises: an acquisition unit 301 for acquiring road video data with timestamps and location information; a preliminary screening unit 302 for automatically identifying traffic signs from the road video data and performing geometric anomaly detection; a consistency determination unit 303 for establishing a time-series tracking sequence for the road signs and confirming whether the road signs have persistent anomalies through multi-frame analysis; a fusion analysis unit 304 for performing local texture consistency and local contour geometric analysis on the road signs to obtain fusion analysis results when persistent anomalies exist; a state evaluation unit 305 for performing state evaluation on the supporting equipment corresponding to the road signs to obtain state evaluation results for the road sign supporting equipment; and a comprehensive determination unit 306 for determining the final surface deformation state and corresponding level of the road signs based on the fusion analysis results and the state evaluation results of the road sign supporting equipment.

[0180] In one embodiment, the consistency determination unit 303 includes:

[0181] The identification maintenance subunit is used to create and maintain a tracking identifier for each road sign by utilizing the position, appearance features, and spatial continuity of the preceding and following frames; the recording subunit is used to record the screening data of each frame in real time based on the tracking identifier to form a time-series tracking sequence; and the analysis subunit is used to analyze the consecutive frames in the time-series tracking sequence using a sliding time window to determine whether the road sign has a persistent anomaly.

[0182] In one embodiment, the primary screening unit 302 includes:

[0183] The identification subunit is used to identify and filter out traffic sign frames that meet the requirements from the road video data, and perform size standardization and edge enhancement processing to obtain a road sign image; the comparison subunit is used to compare the road sign image with the geometric parameters of road signs in the standard model database, and check the consistency of aspect ratio, edge straightness and outer contour shape to perform geometric anomaly detection.

[0184] In one embodiment, the comparison subunit is used to compare the road sign image with the geometric parameters of road signs in a standard model database, compare whether the aspect ratio of the detected object meets the set aspect ratio requirements, evaluate the edge straightness by fitting a straight line equation through boundary detection results, and verify the consistency of the shape by extracting and matching the outer contour.

[0185] In one embodiment, the fusion analysis unit 304 includes:

[0186] The local texture analysis subunit is used to analyze the local texture of the road sign surface to determine whether there are areas of abnormal stretching, twisting, or breakage. If the local texture direction does not meet the set requirements, it is determined that the road sign has potential bending signs, so as to obtain the local texture consistency analysis result. The contour geometry analysis subunit is used to extract the overall contour of the road sign in the continuous frame sequence of the time-tracking sequence and compare the shape changes under different viewpoints. If there are areas with prominent deviations that do not conform to normal perspective transformation, it is determined that the road sign has structural deformation, so as to obtain the contour geometry analysis result. The comprehensive calculation subunit is used to calculate the comprehensive evaluation result based on the local texture consistency analysis result and the contour geometry analysis result, so as to obtain the fusion analysis result.

[0187] In one embodiment, the state evaluation unit 305 includes:

[0188] The equipment extraction subunit is used to locate and extract the supporting equipment corresponding to the road sign based on the road sign location; the comparison subunit is used to calculate the main axis of the supporting equipment using a straight line detection algorithm and compare the main axis of the supporting equipment with a preset vertical direction to obtain a comparison result; the equipment status assessment subunit is used to mark the supporting equipment as abnormal if the comparison result shows that the supporting equipment is tilted, bent or broken, to obtain a status assessment result of the road sign supporting equipment, and to mark the supporting equipment as normal if the comparison result does not show that the supporting equipment is tilted, bent or broken, to obtain a status assessment result of the road sign supporting equipment.

[0189] In one embodiment, the state evaluation unit 305 further includes:

[0190] The reduction subunit is used to reduce the reliability and confidence of the unidentifiable parts of the support device when a portion of the support device is obscured or cannot be clearly identified due to insufficient image resolution.

[0191] In one embodiment, the comprehensive determination unit 306 includes:

[0192] The final state determination subunit is used to determine the final surface deformation state of the road sign based on the fusion analysis results and the state assessment results of the road sign support equipment when the road sign is in an abnormal state; the grade determination subunit is used to determine the grade based on the final surface deformation state of the road sign.

[0193] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the road sign surface deformation detection system 300 based on dynamic video sequences and each unit can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0194] The aforementioned road sign surface deformation detection system 300 based on dynamic video sequences can be implemented as a computer program, which can be used in various ways, such as... Figure 7 It runs on the computer device shown.

[0195] Please see Figure 7 , Figure 7 This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 500 can be a server, wherein the server can be a standalone server or a server cluster composed of multiple servers.

[0196] See Figure 7 The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504.

[0197] The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions that, when executed, cause the processor 502 to perform a method for detecting surface deformation of road signs based on dynamic video sequences.

[0198] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0199] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a road sign surface deformation detection method based on dynamic video sequences.

[0200] This network interface 505 is used for network communication with other devices. Those skilled in the art will understand that... Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 500 to which the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0201] The processor 502 is used to run a computer program 5032 stored in a memory to implement all the steps of the road sign surface deformation detection method based on dynamic video sequences.

[0202] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0203] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0204] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein when executed by a processor, the computer program causes the processor to perform all the steps of the road sign surface deformation detection method based on dynamic video sequences.

[0205] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0206] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0207] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of each unit is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0208] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the system of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0209] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0210] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for detecting surface deformation of a signboard based on a dynamic video sequence, characterized in that, The method comprises the following steps: acquiring road video data with timestamp and location information; automatically identifying traffic signs from the road video data and performing geometric anomaly detection; establishing a time tracking sequence for the traffic sign and determining whether the traffic sign has persistent anomalies through multi-frame analysis; when the traffic sign has persistent anomalies, performing local texture consistency and local contour geometry analysis on the traffic sign to obtain a fusion analysis result; performing state evaluation on a support device corresponding to the traffic sign to obtain a state evaluation result of the traffic sign support device; based on the fusion analysis result and the state evaluation result of the traffic sign support device, determining a final surface deformation state of the traffic sign and a corresponding level; the local texture consistency and local contour geometry analysis on the traffic sign to obtain a fusion analysis result specifically comprises: analyzing the local texture of the traffic sign surface to determine whether there are areas of abnormal stretching, twisting or breaking, and when there is a local texture direction that does not meet the set requirements, it is determined that the traffic sign has potential bending signs to obtain a local texture consistency analysis result; extracting the overall contour of the traffic sign in the continuous frame sequence of the time tracking sequence and comparing the shape changes under different perspectives, and when there is a region with a prominent deviation that does not conform to normal perspective transformation, it is determined that the traffic sign has structural deformation to obtain a contour geometry analysis result; calculating a comprehensive evaluation result according to the local texture consistency analysis result and the contour geometry analysis result to obtain a fusion analysis result.

2. The method of claim 1, wherein, the establishment of a time tracking sequence for the traffic sign and the determination of whether the traffic sign has persistent anomalies through multi-frame analysis comprises: creating and maintaining a tracking identifier for each traffic sign using the location, appearance features and spatial continuity of the front and rear frames of the traffic sign; based on the tracking identifier, recording the screening data of each frame in real time to form a time tracking sequence; using a sliding time window to analyze the continuous frames in the time tracking sequence to determine whether the traffic sign has persistent anomalies.

3. The method of claim 2, wherein the method further comprises: The appearance features include at least one of color distribution and morphological feature vector, and the screening data includes at least one of length-width ratio deviation, outer contour deformation degree and texture anomaly index.

4. The method of claim 1, wherein, the automatic identification of traffic signs from the road video data and the geometric anomaly detection comprises: identifying and screening traffic sign frames that meet the requirements from the road video data, and performing size standardization and edge enhancement processing to obtain traffic sign images; comparing the traffic sign images with the geometric parameters of traffic signs in the standard model database to check the consistency of the length-width ratio, edge straightness and outer contour shape for geometric anomaly detection.

5. The method of claim 4, wherein, the comparison of the traffic sign images with the geometric parameters of traffic signs in the standard model database to check the consistency of the length-width ratio, edge straightness and outer contour shape for geometric anomaly detection comprises: comparing the traffic sign images with the geometric parameters of traffic signs in the standard model database to compare whether the length-width ratio of the detection object meets the set proportion requirement, to evaluate the edge straightness through the straight line equation fitted by the boundary detection result, and to verify the shape consistency through outer contour extraction and matching.

6. The method of claim 1, wherein, The support equipment corresponding to the signboard is evaluated in state to obtain a state evaluation result of the signboard support equipment, including: Positioning and extracting the support equipment corresponding to the signboard based on the signboard; Calculating the main axial direction of the support equipment by using a straight line detection algorithm, and comparing the main axial direction of the support equipment with a preset vertical direction to obtain a comparison result; When the comparison result is that the support equipment has a tilt, bending or breaking phenomenon, the support equipment is marked as an abnormal state to obtain the state evaluation result of the signboard support equipment, and when the comparison result is that the support equipment does not have a tilt, bending or breaking phenomenon, the support equipment is marked as a normal state to obtain the state evaluation result of the signboard support equipment.

7. The method of claim 6, wherein the method further comprises: After the main axial direction of the support equipment is calculated by using the straight line detection algorithm, and the main axial direction of the support equipment is compared with the preset vertical direction to obtain the comparison result, the method further includes: When a part of the support equipment is blocked or cannot be clearly identified due to an image resolution not meeting the requirements, the reliability and confidence of the part of the support equipment that cannot be identified are reduced.

8. The dynamic video sequence based surface deformation detection method of signboard according to claim 1, wherein, Based on the fusion analysis result and the state evaluation result of the signboard support equipment, the final surface deformation state and the corresponding level of the signboard are determined, including: When the signboard is in an abnormal state, the final surface deformation state of the signboard is determined according to the fusion analysis result and the state evaluation result of the signboard support equipment; The level is determined according to the final surface deformation state of the signboard.

9. A sign surface deformation detection system based on dynamic video sequences, characterized in that, It includes: An acquisition unit is configured to acquire road video data with time stamps and location information; An initial screening unit is configured to automatically identify traffic signs from the road video data and perform geometric anomaly detection; A consistency determination unit is configured to establish a time sequence tracking sequence for the signboard and confirm whether the signboard has a persistent anomaly through multi-frame analysis; A fusion analysis unit is configured to perform local texture consistency and local contour geometry analysis on the signboard to obtain a fusion analysis result when the signboard has a persistent anomaly; A state evaluation unit is configured to evaluate the state of the support equipment corresponding to the signboard to obtain a state evaluation result of the signboard support equipment; A comprehensive determination unit is configured to determine the final surface deformation state and the corresponding level of the signboard based on the fusion analysis result and the state evaluation result of the signboard support equipment; The local texture consistency and the local contour geometry analysis on the signboard are performed to obtain a fusion analysis result, specifically including: Analyzing the local texture of the signboard surface to determine whether there is an abnormal stretching, twisting or breaking area, and when there is a local texture direction that does not meet the set requirements, it is determined that the signboard has potential bending signs to obtain a local texture consistency analysis result; The overall contour of the signboard is extracted in the continuous frame sequence of the time sequence tracking sequence, and the shape change under different perspectives is compared, and when there is a region with a prominent deviation that does not conform to normal perspective transformation, it is determined that the signboard has structural deformation to obtain a contour geometry analysis result; An overall evaluation result is calculated according to the local texture consistency analysis result and the contour geometry analysis result, so as to obtain a fusion analysis result.

Images