Manhole cover coding method and system based on dynamic patrol

Abstract

The application discloses a manhole cover coding method and system based on dynamic patrol. The method comprises the following steps: extracting and positioning the manhole cover target of the road surface image, and generating a manhole cover orthographic image; analyzing macroscopic and microscopic attributes based on the manhole cover orthographic image; obtaining a road surface vibration signal, and extracting physical characteristics from the road surface vibration signal by using a dynamics compensation and frequency domain analysis method to obtain a physical tactile feature vector; generating a manhole cover identity code based on environmental changes combined with the macroscopic attributes, the visual feature vector, the spatial topological characteristic value and the physical tactile feature vector; comparing the manhole cover identity code with historical archives in a database to confirm the manhole cover identity or register as a new asset. By implementing the method of the application, the manhole cover asset identity can be accurately locked, high accuracy can be maintained even in the case of environmental changes and complex road conditions, and the efficiency and reliability of urban manhole cover management are significantly improved.

Description

Technical Field

[0001] This invention relates to computer vision, and more specifically to a method and system for coding manhole covers based on dynamic inspection. Background Technology

[0002] The sheer number and striking similarity of manhole covers on urban roads make accurate identification during dynamic inspections a significant challenge. Traditional methods rely primarily on GPS coordinates for identification; however, this approach suffers from drift errors of several meters, especially in urban canyons or areas without archival records, easily leading to "identity swapping" and duplicate filing. Furthermore, for geographically adjacent and extremely similar-looking manhole covers, effective microscopic differentiation methods are lacking; single visual features are easily rendered ineffective by lighting, corrosion, and covering materials, making it difficult to ensure the uniqueness and stability of asset identification.

[0003] Current technologies fail to meet this requirement because they not only rely on unstable GPS absolute coordinates but also ignore the interference of vehicle movement conditions, such as speed and load, on microscopic vibration signals, making it difficult to form a stable and reliable physical fingerprint. Therefore, existing inertial technologies are mainly used for macroscopic navigation and have failed to effectively solve the problem of vibration signal stability at the microscopic level.

[0004] Therefore, it is necessary to design a new method to accurately identify manhole cover assets, maintaining high accuracy even under environmental changes and complex road conditions, and significantly improving the efficiency and reliability of urban manhole cover management. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a manhole cover coding method and system based on dynamic inspection.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a manhole cover coding method based on dynamic inspection, comprising:

[0007] Acquire road surface images captured by vehicle-mounted video capture equipment;

[0008] The manhole cover target is extracted and located from the road surface image, and an orthophoto of the manhole cover is generated;

[0009] Based on the analysis of the orthophoto of the manhole cover, the macroscopic attributes are analyzed, including administrative region, asset category, and filing time;

[0010] Microscopic properties are analyzed based on the orthophoto of the manhole cover, wherein the microscopic properties include visual feature vectors and spatial topological feature values;

[0011] The road surface vibration signal is acquired, and physical features are extracted from the road surface vibration signal using dynamic compensation and frequency domain analysis methods to obtain a physical tactile feature vector.

[0012] Based on environmental changes, a manhole cover identification code is generated by combining the aforementioned macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors.

[0013] The manhole cover identification code is compared with historical records in the database to confirm the manhole cover's identity or register it as a new asset.

[0014] The further technical solution is as follows: The process of acquiring road surface vibration signals and extracting physical features from the road surface vibration signals using dynamic compensation and frequency domain analysis methods to obtain a physical tactile feature vector includes:

[0015] The road vibration signal is acquired, the deviation caused by the vehicle's shock absorption system is removed, and the amplitude is adjusted according to the vehicle speed to obtain the net impact signal. Dynamic gain compensation is then performed to obtain the processed signal.

[0016] The inherent resonant frequency, spectral centroid, and energy decay rate are extracted from the processed signal to obtain physical characteristics characterizing the material properties of the manhole cover.

[0017] The feature binning quantization method is used to map the values ​​of continuous physical features to discrete bucket IDs to obtain physical tactile feature vectors.

[0018] The further technical solution is as follows: The process of acquiring road vibration signals, removing deviations caused by the vehicle's shock absorption system, adjusting the amplitude according to vehicle speed to obtain a net impact signal, and performing dynamic gain compensation to obtain a processed signal includes:

[0019] The road vibration signal is acquired, the mean value of the road background noise in the few seconds before the manhole cover is run over by the vehicle is calculated, and the deviation caused by the vehicle's shock absorption system is removed by the baseline elimination method to obtain the net impact signal.

[0020] The amplitude of the net impact signal is adjusted to the equivalent value at the standard reference speed using a dynamic model based on the real-time GPS vehicle speed to obtain the processed signal.

[0021] The further technical solution is as follows: Extracting the inherent resonant frequency, spectral centroid, and energy attenuation rate from the processed signal to obtain physical characteristics characterizing the material properties of the manhole cover includes:

[0022] The processed signal is converted into a discrete power spectral density, and the sound features characterizing the inherent properties of the manhole cover material are analyzed from it to obtain the physical features characterizing the properties of the manhole cover material. This includes: finding the frequency point with the highest energy in the power spectrum as the inherent resonant frequency; calculating the centroid of the vibration energy distribution in the frequency domain to form the spectral centroid; and evaluating the rate of attenuation of the signal envelope after the impact peak to obtain the energy attenuation rate.

[0023] The further technical solution is as follows: the feature bucketing quantization method maps the values ​​of continuous physical features to discrete bucket IDs to obtain physical tactile feature vectors, including:

[0024] A preset frequency tolerance step size is used to map the values ​​of continuous physical features to discrete bucket IDs to obtain physical tactile feature vectors.

[0025] Its further technical solution is: the microscopic attributes are analyzed based on the orthophoto of the manhole cover, wherein the microscopic attributes include visual feature vectors and spatial topological feature values, including:

[0026] Microscopic key points are extracted from the surface of the manhole cover based on the orthophoto of the manhole cover to generate a visual feature vector; wherein, the microscopic key points include at least one of wear lines, rust spots, and text strokes;

[0027] The curb and lane lines are identified in the road surface image, and the vertical distance from the center of the manhole cover to the curb and the angle between the center of the manhole cover and the lane line are calculated.

[0028] Calculate the path distance between manhole covers at different time intervals;

[0029] Get the grid block ID of the current location;

[0030] The spatial topological feature value is formed by combining the vertical distance from the center of the manhole cover to the curb, the angle between the center of the manhole cover and the lane line, the path distance, and the grid block ID.

[0031] The further technical solution is as follows: the generation of manhole cover identification codes based on environmental changes combined with macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors includes:

[0032] The system performs real-time quality analysis on the visual feature vector, spatial topological feature value, and physical tactile feature vector of the current frame, and dynamically adjusts the weight coefficients of the visual feature vector, spatial topological feature value, and physical tactile feature vector based on the analysis results.

[0033] A weighted fusion strategy is adopted to combine the macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors with corresponding weight coefficients to calculate a comprehensive feature value, which is then converted into an identity identifier to obtain the manhole cover identity code.

[0034] Its further technical solution is: the weight coefficients of the dynamically adjusted visual feature vector, spatial topological feature value, and physical tactile feature vector include:

[0035] When any of the visual feature vector, spatial topological feature value, or physical tactile feature vector becomes unreliable due to environmental factors, the corresponding weight is reduced, and the weights of the remaining features are increased.

[0036] The further technical solution is as follows: comparing the manhole cover identification code with historical records in the database to confirm the manhole cover's identity or register it as a new asset includes:

[0037] Perform a search operation in the database to find all known manhole cover files within a radius of several meters around the current manhole cover, in order to obtain candidate matches;

[0038] Calculate the similarity between the manhole cover identification code and the fingerprint of the candidate matching item. When the similarity exceeds a set threshold and the relative topological location matches, the candidate matching item is a successfully matched object, confirming that the manhole cover is an existing record in the database.

[0039] If multiple inspections fail to find a matching object in the historical archives, but the same unidentified manhole cover exists, mark the manhole cover as a newly added manhole cover, establish an identification code, and add the identification code to the database.

[0040] This invention also provides a manhole cover coding system based on dynamic inspection, comprising:

[0041] The image acquisition unit is used to acquire road surface images captured by the vehicle-mounted video acquisition device;

[0042] The image generation unit is used to extract and locate the manhole cover target from the road surface image and generate an orthophoto image of the manhole cover.

[0043] The analysis unit is used to analyze macroscopic attributes based on the orthophoto of the manhole cover, wherein the macroscopic attributes include administrative region, asset category and filing time;

[0044] The manhole cover differentiation unit is used to analyze microscopic attributes based on the orthophoto of the manhole cover, wherein the microscopic attributes include visual feature vectors and spatial topological feature values;

[0045] The physical feature extraction unit is used to acquire road vibration signals and extract physical features from the road vibration signals using dynamic compensation and frequency domain analysis methods to obtain physical tactile feature vectors.

[0046] The identity generation unit is used to generate a manhole cover identity code based on environmental changes and the macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors.

[0047] The comparison unit is used to compare the manhole cover identification code with historical records in the database to confirm the manhole cover's identity or register it as a new asset.

[0048] The beneficial effects of this invention compared to existing technologies are as follows: This invention acquires road surface images through vehicle-mounted video acquisition equipment, and extracts and locates manhole covers in the images to generate orthophotos. Based on this image analysis, macroscopic attributes including administrative region, asset category, and filing time are analyzed, as well as microscopic attributes such as visual feature vectors and spatial topological feature values. Simultaneously, physical tactile feature vectors are extracted using road vibration signals through dynamic compensation and frequency domain analysis. Combined with environmental changes and the aforementioned multi-dimensional attributes, a unique manhole cover identification code is generated. Finally, this code is compared with historical records in the database to confirm the manhole cover's identity or register it as a new asset. Thus, even under environmental changes and complex road conditions, the identity of manhole cover assets can still be accurately locked, significantly improving the efficiency and reliability of urban manhole cover management.

[0049] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0050] 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.

[0051] Figure 1 A flowchart illustrating the manhole cover coding method based on dynamic inspection provided in an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of a road surface image provided in an embodiment of the present invention;

[0053] Figure 3 A schematic diagram of an orthophoto of a manhole cover provided in an embodiment of the present invention;

[0054] Figure 4 A schematic diagram of relative topological positioning provided in an embodiment of the present invention;

[0055] Figure 5 A schematic diagram of dynamic compensation provided for an embodiment of the present invention;

[0056] Figure 6 This is a schematic diagram of frequency domain fingerprinting provided in an embodiment of the present invention;

[0057] Figure 7 A schematic block diagram of a manhole cover coding system based on dynamic inspection provided in an embodiment of the present invention;

[0058] Figure 8 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0059] 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.

[0060] 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.

[0061] 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.

[0062] 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.

[0063] Please see Figure 1 , Figure 1 This is a flowchart illustrating the dynamic inspection-based manhole cover coding method provided in this embodiment of the invention. This dynamic inspection-based manhole cover coding method is applied in a server. It acquires vehicle-mounted video and road vibration signals through dynamic inspection. Image processing technology is used to extract macroscopic attributes (such as administrative region, asset category, etc.) and microscopic attributes (such as visual feature vectors, spatial topological feature values, etc.) of the manhole cover from the road images. Furthermore, it combines dynamic compensation and frequency domain analysis methods to extract physical tactile feature vectors from the vibration signals. This information is then used to generate a unique manhole cover identification code, which is confirmed or updated by comparing it with historical records in the database. Even under changing environmental conditions and complex road conditions, the system can ensure data accuracy and reliability by adjusting feature weights in real time and eliminating biases, thereby efficiently and accurately managing urban manhole cover assets and significantly improving urban management efficiency and maintenance levels.

[0064] Figure 1This is a flowchart illustrating the manhole cover coding method based on dynamic inspection provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following steps S110 to S170.

[0065] S110. Acquire road surface images captured by the vehicle-mounted video acquisition device.

[0066] In this embodiment, as Figure 2 As shown, road surface images refer to images containing information about the surface of urban roads, dynamically acquired through video capture devices (such as cameras) installed on vehicles. These images not only cover conventional road surface elements such as lanes, curbs, and traffic signs, but also specifically capture and process images of manhole covers.

[0067] Specifically, as the vehicle travels, the onboard video capture equipment continuously captures images of the road surface it passes through, and processes the frames from these video streams as raw data. To ensure accurate identification and feature extraction of manhole covers, the system first preprocesses these raw road surface images. This includes, but is not limited to:

[0068] ROI Extraction: The system activates the Region of Interest (ROI) extraction module, automatically locating and cropping out the parts of the image that may contain manhole covers (circular or square). This process effectively reduces unnecessary background information interference, improving the speed and accuracy of subsequent processing.

[0069] Perspective Transformation: Since changes in shooting angle can cause geometric distortions in manhole cover images, perspective transformation techniques are needed to correct the selected Region of Interest (ROI). This eliminates distortions caused by different shooting angles and outputs a standardized "manhole cover orthophoto." This orthophoto is an idealized representation of the manhole cover viewed from a vertically overhead perspective, which is crucial for ensuring the consistency and reliability of feature extraction.

[0070] Feature extraction preparation: The generated orthophoto of the manhole cover lays the foundation for subsequent high-precision feature extraction. This includes the analysis of macroscopic attributes (such as administrative region, asset category, and filing time) and microscopic attributes (such as visual feature vectors and spatial topological feature values), as well as the construction of physical tactile feature vectors based on vibration signals. These steps provide a clear and distortion-free working object.

[0071] In summary, acquiring road surface images from vehicle-mounted video capture equipment is not only the first step in the entire manhole cover coding process, but also a crucial one. It directly affects the data quality of all subsequent processing steps and the accuracy and uniqueness of the final manhole cover identification code. Using the methods described above, effective management and tracking of manhole cover assets can be achieved even in the face of complex and ever-changing urban road conditions and environmental circumstances.

[0072] S120. Extract and locate the manhole cover target from the road surface image, and generate an orthophoto of the manhole cover.

[0073] In this embodiment, the manhole cover orthophoto refers to a standardized image generated by processing road surface images acquired by vehicle-mounted video acquisition equipment. This image displays the manhole cover from a vertically upward perspective, eliminating geometric distortions caused by changes in the shooting angle, thereby ensuring the accuracy and consistency of subsequent feature extraction and recognition processes.

[0074] Specifically, the process of generating orthophotos of manhole covers includes the following key steps:

[0075] First, video capture devices, such as cameras mounted on patrol vehicles, capture real-time images of the city's road surface. These images include not only standard elements like lanes and curbs, but also multiple views of the target manhole cover.

[0076] The system automatically locates and crops the parts of the image that may contain manhole covers (such as circles or squares), a process called ROI extraction. This step effectively reduces unnecessary background information interference and allows resources to be focused on further processing of the region of interest.

[0077] Because changes in the shooting angle during vehicle movement can cause geometric distortions in manhole cover images, perspective transformation techniques are needed to correct the selected Region of Interest (ROI). Perspective transformation is a mathematical method that maps points on a two-dimensional plane to another plane to correct image distortion caused by differences in viewing angle. This method outputs an idealized representation of the manhole cover viewed from a vertically upward perspective—a "manhole cover orthophoto." This not only improves the consistency of feature extraction but also lays the foundation for subsequent high-precision analysis.

[0078] The generated orthophoto of the manhole cover provides a clear working object for subsequent microscopic differentiation schemes, such as visual fingerprint extraction (extracting microscopic key points such as wear patterns and rust spots) and relative topological positioning (calculating the normal distance from the center of the manhole cover to the curb and the angle between its orientation and the lane line). In addition, the orthophoto also supports the construction of physical tactile feature vectors based on vibration signals, as well as an adaptive identity locking mechanism in all-weather environments.

[0079] In summary, in this embodiment, as Figure 3 As shown, the "manhole cover orthophoto" is obtained through a series of complex image processing techniques, aiming to provide a distortion-free, standardized view of the manhole cover. This is crucial for accurately identifying assets that are resistant to drift, duplication, and environmental changes. In this way, even in the face of complex and ever-changing urban road conditions and environmental conditions, the identity information of each manhole cover can be effectively managed and tracked.

[0080] like Figure 3 As shown, after optimizing the features extracted by the image processing algorithm, the differences and directions between standard features and differential features are clarified.

[0081] Standard features (such as ORB or SIFT) refer to microscopic dot-shaped features extracted from manhole cover images using image processing techniques. These features are generic and used to identify the basic patterns and structures of manhole covers.

[0082] In contrast, the difference features specifically refer to the unique macroscopic blocky physical differences that the manhole cover exhibits compared to its original design state due to factors such as corrosion and wear. To more accurately identify the manhole cover, we clearly marked the most significant areas of macroscopic physical damage—those irregular, large areas of rust—with three red circles on the image. These areas represent the manhole cover's unique "macroscopic fingerprint feature blocks," which are difficult to forge or replicate, thus becoming key evidence for confirming the manhole cover's identity.

[0083] By highlighting and marking these unique macroscopic feature blocks, the ability to distinguish similar manhole covers can be significantly improved, creating a more stable and reliable physical fingerprint system. This method not only enhances the identifiability of individual manhole covers but also provides a solid foundation for manhole cover management and tracking.

[0084] like Figure 3 As shown, the system finally quantifies the distribution of the identified micro-key points and the macro-corrosion and wear areas to generate a unique visual feature vector for the manhole cover. Then, the "multidimensional differential hash algorithm" is used to compress the high-dimensional feature vector and convert it into a fixed-length hash string, such as "A7F2".

[0085] S130. Analyze macroscopic attributes based on the orthophoto of the manhole cover, wherein the macroscopic attributes include administrative region, asset category, and filing time.

[0086] In this embodiment, macro-attributes refer to the set of features that provide basic information about manhole covers but do not directly distinguish specific individuals. These attributes are mainly used for rapid indexing and classification management, facilitating efficient identification and tracking of large-scale assets. Specifically, macro-attributes include, but are not limited to, key information such as administrative region, asset category, and filing time.

[0087] Administrative Region: The geographic coordinates of the current manhole cover are obtained using GPS positioning technology and then matched against a national administrative division database. A corresponding numerical code (e.g., "310115") is generated, representing the specific administrative region where the manhole cover is located. This coding method helps to quickly determine the approximate location of the manhole cover and facilitates management and retrieval by region.

[0088] Asset Category (Type): Deep learning algorithms are used to analyze and identify images or video streams of manhole covers. Based on characteristics such as shape and material, the covers are categorized into different types. Short letter codes are used to represent different types, such as "MH" for round metal manhole covers and "RG" for rectangular storm drain grates. This simplifies the data storage structure and facilitates subsequent data processing and querying.

[0089] Creation Date: Extracted directly from the timestamps of the data collected by the system, ensuring that each manhole cover has a clear creation record date. It is typically in YYYYMMDD format (e.g., "20251024"), accurate to the day. This clearly reflects the specific date the manhole cover was first entered into the system, which is crucial for tracking changes throughout its lifecycle.

[0090] Specifically, the GPS coordinates in the orthophoto are matched to the corresponding administrative division codes. The morphological characteristics of the manhole covers are intelligently analyzed to determine their category and assign a corresponding code. The system time at the time of data collection is used as the filing date of the manhole cover, forming a complete macroscopic attribute description.

[0091] In summary, in this embodiment, "analyzing macroscopic attributes based on the orthophoto of the manhole cover" refers to processing the manhole cover image to extract a series of feature parameters reflecting its basic identity information, including but not limited to the administrative region, asset category, and filing time. This information collectively constitutes the basic layer of the manhole cover's identity identification, providing the necessary prerequisites for further microscopic differentiation, physical fingerprint construction, and full lifecycle management.

[0092] S140. Analyze the microscopic properties based on the orthophoto of the manhole cover, wherein the microscopic properties include visual feature vectors and spatial topological feature values.

[0093] In this embodiment, micro-attributes refer to the set of detailed features that can be used to accurately distinguish different individual manhole covers. These features not only help solve the problem of identifying geographically adjacent and similar-looking manhole covers, but also provide a unique identifier for each manhole cover. Specifically, micro-attributes include visual feature vectors and spatial topological feature values.

[0094] In one embodiment, step S140 described above may include steps S141 to S145.

[0095] S141. Based on the orthophoto of the manhole cover, extract microscopic key points from the surface of the manhole cover to generate a visual feature vector; wherein, the microscopic key points include at least one of wear lines, rust spots, and text strokes.

[0096] In this embodiment, the microscopic key points refer to the detailed features on the surface of the manhole cover, such as wear lines, rust spots, and strokes of text. These subtle but stable features can be effectively extracted using ORB (Oriented Fast and Rotated BRIEF) or SIFT (Scale-Invariant Feature Transform) algorithms.

[0097] By analyzing and quantifying the aforementioned microscopic key points, a vector describing the unique surface characteristics of the manhole cover is formed. Even for manhole covers manufactured in the same batch, their visual feature vectors will differ due to natural wear and tear and environmental influences during use, thus enabling accurate differentiation.

[0098] S142. Identify the curb and lane lines in the road surface image, and calculate the vertical distance from the center of the manhole cover to the curb and the angle between the center of the manhole cover and the lane line.

[0099] In this embodiment, firstly, the positions of the curbstone and lane lines are identified in the image. Then, the vertical distance from the center of the manhole cover to the curbstone (denoted as ) is calculated. ) and the angle between the center of the manhole cover and the lane line (denoted as ) These two parameters together constitute part of the spatial location information of the manhole cover relative to its surrounding environment, which helps to provide additional positioning accuracy when GPS signals are unstable.

[0100] S143. Calculate the path distance between manhole covers at different time sequences.

[0101] When the current manhole cover (Node N) is detected, the system automatically traces back the position information of the previous manhole cover (Node N-1) and calculates the precise path distance between them using a visual odometer or a high-precision wheel speed meter (denoted as ). This method can maintain tracking of the manhole cover's position by accumulating relative position changes during continuous monitoring, thus enhancing the stability of the positioning.

[0102] S144. Obtain the grid block ID of the current location;

[0103] S145. Combine the vertical distance from the center of the manhole cover to the curb, the angle between the center of the manhole cover and the lane line, the path distance, and the grid block ID to form a spatial topological feature value.

[0104] In this embodiment, spatial topological feature value refers to a set of data that reflects the unique geometric position and relative relationship of the manhole cover in the pipeline chain by integrating the vertical distance from the center of the manhole cover to the curb, the angle with the lane line, the path distance relative to the previous manhole cover, and the ID of the grid block in which it is located.

[0105] Finally, all the information obtained above—the vertical distance from the center of the manhole cover to the curb—is used to determine the optimal distance. The angle between the center of the manhole cover and the lane line Path distance And the grid block ID within a 10m radius of GPS (denoted as...) —Integrated together, this forms a spatial topological characteristic value that comprehensively reflects the unique "geometric coordinates" of the manhole cover within its respective pipeline network chain. This data not only considers the absolute position of the manhole cover itself, but also its relative relationship with other manhole covers and the surrounding environment, ensuring that the identity of each manhole cover can be accurately identified even in complex urban environments.

[0106] In summary, this embodiment involves a series of sophisticated operational procedures for "analyzing microscopic attributes based on the orthophoto of the manhole cover," including extracting microscopic key points from the manhole cover surface to generate visual feature vectors and constructing spatial topological feature values ​​by combining environmental factors. These measures work together to overcome the problems encountered by traditional methods when dealing with geographically adjacent manhole covers that are highly similar in appearance, thereby improving the accuracy and reliability of asset identification.

[0107] The aforementioned step S140 primarily addresses the scenario in urban environments where there are numerous twin manhole covers that are geographically very close (only a few meters apart) and have highly similar appearances. Because traditional GPS technology has a positioning error of 3-5 meters, and the single visual characteristics (such as texture) of industrially standardized products are almost identical, this leads to frequent problems of manhole cover identity swapping or coding conflicts.

[0108] like Figure 4 As shown, to address this issue, a dual anchoring mechanism combining visual and relative topology is proposed. This system does not rely solely on absolute coordinates, but rather utilizes the relative geometric invariants between the manhole cover and its surrounding environment for differentiation:

[0109] Visual fingerprint extraction (texture micro-distinction): Using algorithms such as ORB / SIFT, unique wear patterns, rust spots, and text strokes on the surface of the manhole cover are extracted to generate a unique visual feature vector. In this way, even manhole covers from the same batch can be distinguished by differences in wear and tear during use.

[0110] Relative topological positioning (spatial micro-distinction): In order to accurately distinguish adjacent manhole covers when GPS signals are weak or insufficient in accuracy, this system constructs a topological coordinate system including two dimensions: horizontal and vertical.

[0111] Lateral Anchor: Calculates the distance from the center of the manhole cover to the curb by identifying the "curbstone" and "lane lines" in the image. ) and the angle between its direction and the lane line ( If these parameters of two manhole covers are different, they can be accurately distinguished.

[0112] Longitudinal Chain Constraint: The system maintains a dynamic tracking queue. When the current manhole cover is detected, it automatically traces back the position information of the previous manhole cover and calculates the precise path distance between the two using a visual odometer or a high-precision wheel speed meter. ).

[0113] Finally, these elements are integrated into a spatial topological eigenvalue containing dual constraints. ,in It is the grid block ID within a radius of 10 meters. This relationship based on relative position constitutes the unique "geometric coordinates" of each manhole cover in its pipeline chain.

[0114] Specifically, such as Figure 4 As shown, the precise definition of the location of a manhole cover includes the following aspects:

[0115] Normal distance d: This refers to the vertical distance from the center of the manhole cover to the curb. As a "lateral anchor point," it helps to accurately determine the position of the manhole cover in the transverse direction of the lane.

[0116] Deflection angle θ: refers to the angle between the center of the manhole cover and the lane line, which reflects the orientation of the manhole cover relative to the road direction.

[0117] Chain distance ΔL: Represents the precise path distance between the current manhole cover and the previous manhole cover. This parameter acts as a "vertical chain constraint," connecting scattered manhole covers into a logical chain.

[0118] The method for calculating these parameters is as follows:

[0119] The system first identifies the curb and lane lines in the road image, then uses geometric calculations to derive the distance *d* and angle *θ* from the center of the manhole cover to these reference points. For the chain distance ΔL, GPS is not used for calculation due to potential drift errors. Instead, visual odometry or high-precision wheel speedometers are employed to accumulate and calculate the precise path length during vehicle travel. Finally, these geometric parameters are combined with the grid block ID of the current location to form a unique spatial topological feature value *S* for the manhole cover. topo。

[0120] Furthermore, the spatial topological eigenvalue S topo。Further transformation using a multidimensional differential hash algorithm generates a fixed-length hash string, such as "A7F2".

[0121] This method not only improves the accuracy of manhole cover location description but also provides strong support for manhole cover management, tracking, and maintenance. In this way, each manhole cover can be uniquely and accurately located, greatly facilitating subsequent related work.

[0122] S150. Acquire road vibration signals and extract physical features from the road vibration signals using dynamic compensation and frequency domain analysis methods to obtain physical tactile feature vectors.

[0123] In this embodiment, the physical tactile feature vector refers to the feature set that is used to extract the inherent resonant frequency, spectral centroid, and energy attenuation rate that characterize the material properties of the manhole cover by analyzing the road vibration signal during vehicle travel, and mapping them to discrete bucket IDs to form a unique identifier for the manhole cover.

[0124] In one embodiment, step S150 described above may include steps S151 to S153.

[0125] S151. Acquire road vibration signals, remove deviations caused by the vehicle's shock absorption system, adjust the amplitude according to vehicle speed to obtain a net impact signal, and perform dynamic gain compensation to obtain a processed signal.

[0126] In one embodiment, step S151 described above may include steps S1511 to S1512.

[0127] S1511. Obtain road vibration signals, calculate the average road background noise several seconds before the manhole cover is run over by a vehicle, and remove the deviation caused by the vehicle's shock absorption system using the baseline elimination method to obtain the net impact signal.

[0128] S1512. The amplitude of the net impact signal is adjusted to the equivalent value under the standard reference speed using a dynamic model based on the real-time GPS vehicle speed, so as to obtain the processed signal.

[0129] In this embodiment, the system first collects road vibration signals in real time using sensors installed on the vehicle. Since these signals may be affected by the vehicle's suspension system and speed changes, a series of preprocessing operations are required to eliminate these interfering factors:

[0130] Remove deviation: before calculating the gland Average data per second ( This serves as the background noise of the road surface, used for baseline elimination to obtain the net impact signal. .

[0131] Dynamics gain compensation: combined with real-time GPS vehicle speed ( ) and standard reference speed ( The amplitude is adjusted using a dynamic model, and gain compensation is performed to obtain the processed signal. .

[0132] S152. Extract the inherent resonant frequency, spectral centroid, and energy attenuation rate from the processed signal to obtain physical characteristics that characterize the material properties of the manhole cover.

[0133] In this embodiment, physical characteristics refer to key parameters extracted from vibration signals that reflect the material properties of the manhole cover, such as the inherent resonant frequency, the centroid of energy distribution in the frequency domain, and energy attenuation characteristics.

[0134] Specifically, the processed signal is converted into a discrete power spectral density, and the sound features characterizing the inherent properties of the manhole cover material are analyzed from it to obtain the physical features characterizing the properties of the manhole cover material; this includes: finding the frequency point with the highest energy in the power spectrum as the inherent resonant frequency; calculating the centroid of the vibration energy distribution in the frequency domain to form the spectral centroid; and evaluating the rate of attenuation of the signal envelope after the impact peak to obtain the energy attenuation rate.

[0135] Next, frequency domain analysis is performed on the processed signal to extract three main physical characteristics that characterize the material properties of the manhole cover:

[0136] Inherent resonant frequency By performing a short-time Fourier transform (STFT) on the processed signal, the time-domain waveform is converted into a discrete power spectral density P(fk), and the frequency point with the highest energy is identified as the inherent resonant frequency of the manhole cover. .

[0137] Spectral centroid This involves calculating the centroid of the vibration energy distribution in the frequency domain, reflecting the overall damping characteristics of the manhole cover material. The specific formula is as follows: , where N represents the total number of frequency points within the effective frequency band.

[0138] Energy decay rate The decay rate of the signal envelope after the peak impact is evaluated to characterize the energy absorption properties of the manhole cover material.

[0139] S153. The feature bucket quantization method is used to map the values ​​of continuous physical features to discrete bucket IDs to obtain physical tactile feature vectors.

[0140] In this embodiment, a preset frequency tolerance step size is used to map the values ​​of continuous physical features to discrete bucket IDs in order to obtain physical tactile feature vectors.

[0141] To ensure the generation of stable physical tactile feature vectors even with minute measurement errors, the system employs a feature binning quantization method:

[0142] Set tolerance step size: Preset a frequency tolerance step size (e.g., 5Hz) to define the width of the "bucket".

[0143] Mapping to Bucket ID: Mapping consecutive physical feature values ​​(such as...) Map it to the corresponding bucket ID based on its size.

[0144] Generating Feature Vectors: The final physical tactile feature vector contains information in three dimensions: intrinsic resonant frequency, spectral centroid, and energy decay rate. Quantization ensures consistency even for data collected at different times or under different conditions. .

[0145] This series of steps effectively solves the problem of unstable vibration signals caused by environmental changes and vehicle conditions, thereby achieving effective identification of the manhole cover.

[0146] In this embodiment, step S150 addresses the following scenario: Existing inertial technology is mainly used for navigation. When directly applied to manhole cover recognition, the vibration signal is significantly affected by vehicle driving conditions (such as speed and load) and road surface conditions (such as smoothness). For example, the vibration amplitude generated by the same manhole cover at different vehicle speeds (20km / h vs. 60km / h) may vary greatly, making it difficult to form a stable physical fingerprint for recognition.

[0147] Therefore, this embodiment adopts the following... Figure 5 and Figure 6 The dynamic compensation and frequency domain fingerprinting shown are used to overcome the above challenges and extract the inherent physical characteristics of the manhole cover. The "deep quantization" process is proposed to eliminate the influence of environmental variables.

[0148] Specifically, Figure 5 and Figure 6 These are two different perspectives on the same set of data. Figure 5 It shows the chaotic state on the surface of the data, while Figure 6 Mathematical tools were used to "see through" the internal structure of the data, and key physical features hidden in the noise (vibration at 142.5 Hz) were found and ultimately used to generate unique digital identifiers.

[0149] Specifically, first cut the section before the cap is pressed. Calculate the average background noise of the road surface using data from the second. Baseline elimination is performed to remove systematic deviations from the vehicle's shock absorption system, resulting in a net impact signal. Subsequently, using real-time GPS vehicle speed ( The amplitude is adjusted to the standard reference velocity using a dynamic model. The equivalent value under () is used to perform dynamic gain compensation: .

[0150] For normalized signals Performing a short-time Fourier transform (STFT) converts the time-domain waveform into a discrete power spectral density in the frequency domain. (in For the first One frequency component, (This corresponds to the energy amplitude). From this, the inherent acoustic signature characteristics of the material can be calculated:

[0151] Inherent resonant frequency ( Find the frequency point with the highest energy in the power spectrum to characterize the principal modal stiffness of the manhole cover. .

[0152] Spectral centroid ( Source and Definition: This variable represents the centroid of vibration energy distribution in the frequency domain (similar to the center of mass in physics), reflecting the overall damping characteristics of the manhole cover material (cast iron has a crisp tone, cement has a dull tone). Calculation Process: The system uses frequency... As variables, with the corresponding power spectral amplitude Calculate the weighted average using the weights: (in (Total number of frequency points within the effective frequency band);

[0153] Energy decay rate ( ): Calculate the signal envelope decay rate after the peak impulse.

[0154] The final generated physical tactile feature vector contains the above three dimensions: .

[0155] To address the minute measurement errors that may occur during continuous frame observations or multiple inspections, a feature-based bucket quantization mechanism is introduced to prevent the generation of redundant codes, as detailed below:

[0156] Setting tolerance bucket: The system presets the frequency tolerance step size. (e.g., 5Hz).

[0157] Discretization mapping: Mapping continuous feature values ​​to discrete "bucket IDs": .

[0158] Uniqueness Guarantee: Even Fluctuations within a small range, as long as they fall within the same tolerance range, the quantification result... Maintaining consistency. The system uses hash deduplication logic to ensure that only one unique identification code is generated for the same physical manhole cover.

[0159] S160. Based on environmental changes, combine the macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors to generate manhole cover identification codes.

[0160] In this embodiment, the manhole cover identification code refers to a unique identifier calculated by integrating macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors, and dynamically adjusting the weights of each feature according to environmental conditions, using a weighted fusion strategy. This identifier is used to accurately identify and manage the entire lifecycle information of manhole covers.

[0161] This step aims to integrate multi-source data, including macro-level attributes (such as administrative region, asset category, and filing time) and visual feature vectors ( Spatial topological eigenvalues ​​( ) and physical tactile feature vectors ( A stable and unique manhole cover identification code is dynamically generated, taking into account environmental changes.

[0162] In one embodiment, step S160 described above may include steps S161 to S162.

[0163] S161. Perform real-time quality analysis on the visual feature vector, spatial topological feature value, and physical tactile feature vector of the current frame, and dynamically adjust the weight coefficients of the visual feature vector, spatial topological feature value, and physical tactile feature vector based on the analysis results.

[0164] In this embodiment, when any of the visual feature vector, spatial topological feature value, and physical tactile feature vector becomes unreliable due to interference from environmental factors, the corresponding weight is reduced, and the weights of the remaining features are increased.

[0165] Visual feature vectors ( ): Microscopic key points (wear lines, rust spots, etc.) on the surface of the manhole cover are extracted using algorithms such as ORB / SIFT. When lighting conditions are poor or the manhole cover is covered, visual features may become unreliable.

[0166] Spatial topological eigenvalues ​​( This is a dual anchoring mechanism based on the relative position of the manhole cover to its surroundings (such as curbs and lane markings). The effectiveness of this feature may decrease when the curb is obstructed or the GPS signal is unstable.

[0167] Physical tactile feature vector ( The inherent acoustic signature of the manhole cover is obtained by performing dynamic compensation and frequency domain fingerprinting on the vibration signals generated during vehicle movement. Although it is less affected by the vehicle's condition, it may be affected in extreme environments (such as extremely uneven road surfaces).

[0168] The system will evaluate the quality of each feature in real time. When a feature becomes unreliable due to environmental factors, its corresponding weight coefficient (α, β, γ) will be reduced, while the weights of other more reliable features will be increased to ensure the accuracy and uniqueness of the final generated identity code.

[0169] S162. A weighted fusion strategy is adopted to combine the macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors with corresponding weight coefficients to calculate a comprehensive feature value, which is then converted into an identity identifier to obtain the manhole cover identity code.

[0170] In this step, the system will use the formula This is used to calculate the comprehensive feature value of the manhole cover and generate the final identification code. Specifically:

[0171] Macro numerical prefix ( The index consists of the administrative region code where the manhole cover is located, the asset type code, and the filing date, providing a basis for a quick index.

[0172] Weighted fusion strategy: Based on the weight coefficients (α, β, γ) determined in step S161, the visual feature vector, spatial topological feature value and physical tactile feature vector are weighted and summed, and then converted into a fixed-length string as an identity identifier through a hash function.

[0173] This environmentally adaptive adjustment strategy not only solves the limitations of traditional single identification methods (such as GPS coordinates and visual features), but also greatly improves the accuracy and reliability of manhole cover identification. Especially in complex urban environments, it can effectively cope with various challenges and achieve "zero-manual" automated filing and full life cycle management.

[0174] Step S160 above addresses a scenario where, during actual inspections, relying on a single feature for manhole cover identification is easily affected by environmental factors and may fail. For example, visual features may not function properly at night or when the cover is covered in mud; obstructions from tall buildings may cause GPS signal drift, affecting positioning accuracy; and roadside parking may obstruct curbs, causing spatial topological features to fail.

[0175] To address the aforementioned issues, a technical solution called Dynamic-ID is proposed, which generates a unique identifier (UID) through adaptive fusion.

[0176] The system can evaluate the data quality of the current frame in real time and dynamically adjust the weighting coefficients α, β, and γ accordingly. Its working principle is as follows:

[0177] When visual features are obstructed (such as insufficient lighting or surface covering), the system automatically reduces the visual feature weight α while increasing the physical tactile feature weight γ, relying on vibration sound and relative position information to determine the identity of the manhole cover.

[0178] When the location features are obstructed (e.g., the curb is blocked), the system will reduce the spatial topological feature weight β and instead rely on visual texture features and physical tactile features to identify the manhole cover.

[0179] This method ensures accurate and reliable identification and management of manhole covers, even in complex and variable environments.

[0180] S170. Compare the manhole cover identification code with the historical records in the database to confirm the manhole cover's identity or register it as a new asset.

[0181] In this crucial step, the system compares the generated manhole cover identification code (Dynamic-ID) with historical records in the database to confirm the manhole cover's identity or register it as a new asset.

[0182] In one embodiment, step S170 described above may include steps S171 to S173.

[0183] S171. Perform a search operation in the database to find all known manhole cover files within a radius of several meters around the current manhole cover, in order to obtain candidate matches.

[0184] In this embodiment, candidate matching items refer to a list of known manhole cover files that may match the current manhole cover, filtered out during the database retrieval operation based on the current manhole cover's location and its multi-dimensional features (visual, spatial topology, physical tactile).

[0185] First, the system performs a search in the database to find all known manhole cover files within a certain radius around the current manhole cover. This radius is typically set to a radius of several meters (e.g., 10 meters) because multiple manhole covers with similar appearances or close locations may exist within this range. By limiting the search area, the number of candidate matches can be effectively reduced, improving the efficiency and accuracy of subsequent processing.

[0186] Search criteria: In addition to geographic location information, the system also utilizes visual feature vectors. Spatial topological features and physical tactile feature vector Used as search criteria.

[0187] The result of this step is a list of candidate matches that may be associated with the current manhole cover.

[0188] S172. Calculate the similarity between the manhole cover identification code and the fingerprint of the candidate matching item. When the similarity exceeds a set threshold and the relative topological position matches, the candidate matching item is a successfully matched object, and the manhole cover is confirmed to be an existing record in the database.

[0189] Next, the system calculates the similarity between the current manhole cover's identification code (Dynamic-ID) and each candidate match. This step involves comparing multi-dimensional features:

[0190] Visual feature similarity: Visual fingerprints extracted using algorithms such as ORB / SIFT are compared with detailed features such as wear patterns and rust spots on the surface of the manhole cover.

[0191] Spatial topological similarity: Check whether the positional relationship between the manhole cover and environmental elements (such as curb stones, lane lines) is consistent.

[0192] Physical tactile feature similarity: Analysis of vibration acoustic signature features, including the inherent resonant frequency. Energy decay rate Spectral centroid Consistency of parameters.

[0193] When the similarity of each of the above dimensions exceeds the set threshold (e.g., 85%), and the relative topological location also matches, the system considers that a matching object has been found and confirms that the manhole cover belongs to an existing record in the database.

[0194] S173. If multiple inspections fail to find a matching object in the historical archives, but the same unidentified manhole cover exists, mark the manhole cover as a newly added manhole cover, establish an identification code, and add the identification code to the database.

[0195] If, during multiple inspections, a matching manhole cover cannot be found in the historical archives, but an unidentified manhole cover is discovered, the system will automatically mark it as a newly added manhole cover. At this point, the system needs to create a unique identification code for this new manhole cover and add it to the database for future tracking and management.

[0196] Ensuring uniqueness: Through hash deduplication logic and feature bucketing quantization mechanism, the global uniqueness and anti-duplication properties of the identity code are maintained even under small measurement errors.

[0197] Automated process: The entire process requires no human intervention, achieving the goal of "zero human intervention" automated record-keeping and full lifecycle management.

[0198] In summary, through these three steps, this embodiment not only solves the problem of confusion between adjacent "twin manhole covers", but also breaks through the limitations of traditional visual recognition, ensuring the global uniqueness and anti-duplication of asset codes, while realizing efficient and accurate dynamic inspection and management of manhole covers.

[0199] This embodiment aims to solve the technical challenge of accurately distinguishing "twin assets" and maintaining their identity throughout their entire lifecycle during dynamic inspections of urban manhole covers. It achieves precise identification of assets that are resistant to drift, duplication, and environmental changes.

[0200] This system integrates data from multiple sensors (such as vision, inertial, and GPS) to provide a comprehensive description of the location and characteristics of manhole covers. Addressing the challenge of "geographically adjacent and similar-looking" covers, it enhances recognition accuracy by analyzing the local topology around the manhole cover and using a multi-dimensional differential hashing algorithm, effectively distinguishing manhole covers with extremely similar appearances. To address "vehicle and environmental interference," signal processing techniques are employed to eliminate or reduce the impact of vehicle movement on microscopic vibration signals, ensuring the stability of the physical fingerprint. By combining time series analysis and machine learning algorithms, the system updates the status information of the manhole cover in real time, guaranteeing the accuracy and uniqueness of its identity throughout its entire lifecycle.

[0201] Therefore, the method in this embodiment completely solves the problem of confusion between adjacent "twin manhole covers," accurately distinguishing them through a unique local topological structure and a multi-dimensional differential hash algorithm. It overcomes the environmental limitations and blind spots of visual recognition, utilizing data from multiple sensors and advanced signal processing techniques to improve recognition accuracy. It ensures the global uniqueness and anti-duplication properties of asset codes, employing innovative coding strategies to avoid coding conflicts. It achieves "zero-manual" automated filing and full lifecycle management, greatly improving the efficiency and reliability of urban management.

[0202] In summary, the method of this embodiment not only solves many defects in traditional technologies, but also provides a brand-new solution for the management and maintenance of urban manhole covers, which has important practical application value and broad application prospects.

[0203] The aforementioned manhole cover coding method based on dynamic inspection acquires road surface images through vehicle-mounted video acquisition equipment, extracts and locates manhole covers in the images to generate orthophotos, and analyzes these images to obtain macroscopic attributes such as administrative region, asset category, and filing time, as well as microscopic attributes such as visual feature vectors and spatial topological feature values. Simultaneously, it utilizes road vibration signals and extracts physical tactile feature vectors through dynamic compensation and frequency domain analysis, combining environmental changes and the aforementioned multi-dimensional attributes to generate a unique manhole cover identification code. Finally, this code is compared with historical records in the database to confirm the manhole cover's identity or register it as a new asset. This allows for accurate identification of manhole cover assets even under environmental changes and complex road conditions, significantly improving the efficiency and reliability of urban manhole cover management.

[0204] Figure 7 This is a schematic block diagram of a manhole cover coding system 300 based on dynamic inspection provided in an embodiment of the present invention. Figure 7 As shown, corresponding to the above-described manhole cover coding method based on dynamic inspection, the present invention also provides a manhole cover coding system 300 based on dynamic inspection. This system includes a unit for executing the above-described manhole cover coding method based on dynamic inspection, which can be configured in a server. Specifically, please refer to... Figure 7 The system includes an image acquisition unit 301, an image generation unit 302, an analysis unit 303, a manhole cover differentiation unit 304, a physical feature extraction unit 305, an identity generation unit 306, and a comparison unit 307.

[0205] Image acquisition unit 301 is used to acquire road surface images captured by vehicle-mounted video acquisition equipment; image generation unit 302 is used to extract and locate manhole cover targets from the road surface images and generate manhole cover orthophotos; analysis unit 303 is used to analyze macroscopic attributes based on the manhole cover orthophotos, wherein the macroscopic attributes include administrative region, asset category, and filing time; manhole cover differentiation unit 304 is used to analyze microscopic attributes based on the manhole cover orthophotos, wherein the microscopic attributes include visual feature vectors and spatial topological feature values; physical feature extraction unit 305 is used to acquire road surface vibration signals and extract physical features from the road surface vibration signals using dynamic compensation and frequency domain analysis methods to obtain physical tactile feature vectors; identity generation unit 306 is used to generate manhole cover identity codes based on environmental changes combined with the macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors; comparison unit 307 is used to compare the manhole cover identity codes with historical files in the database to confirm the manhole cover identity or register it as a new asset.

[0206] In one embodiment, the physical feature extraction unit 305 includes:

[0207] The vibration processing subunit is used to acquire road vibration signals, remove deviations caused by the vehicle's shock absorption system, adjust the amplitude according to vehicle speed to obtain a net impact signal, and perform dynamic gain compensation to obtain a processed signal; the extraction subunit is used to extract the inherent resonant frequency, spectral centroid, and energy attenuation rate from the processed signal to obtain physical characteristics characterizing the material properties of the manhole cover; the mapping subunit is used to map the values ​​of continuous physical features to discrete bucket IDs using a feature bucket quantization method to obtain a physical tactile feature vector.

[0208] In one embodiment, the vibration processing subunit includes: a calculation module for acquiring road vibration signals, calculating the average road background noise several seconds before the manhole cover is run over by a vehicle, and removing the deviation caused by the vehicle's shock absorption system using a baseline elimination method to obtain a net impact signal; and an adjustment module for adjusting the amplitude of the net impact signal to an equivalent value at a standard reference speed based on the real-time GPS vehicle speed using a dynamic model to obtain the processed signal.

[0209] In one embodiment, the extraction subunit is used to convert the processed signal into a discrete power spectral density and analyze the sound features characterizing the inherent properties of the manhole cover material to obtain physical features characterizing the properties of the manhole cover material; wherein, the extraction subunit includes: finding the frequency point with the highest energy in the power spectrum as the inherent resonant frequency; calculating the centroid of the vibration energy distribution in the frequency domain to form the spectral centroid; and evaluating the rate of attenuation of the signal envelope after the impact peak to obtain the energy attenuation rate.

[0210] In one embodiment, the mapping subunit is used to preset a frequency tolerance step size and map the values ​​of continuous physical features to discrete bucket IDs to obtain a physical tactile feature vector.

[0211] In one embodiment, the manhole cover distinguishing unit 304 includes:

[0212] The key point extraction subunit is used to extract microscopic key points from the surface of the manhole cover based on the orthophoto of the manhole cover to generate a visual feature vector; wherein, the microscopic key points include at least one of wear texture, rust spots, and text strokes; the numerical calculation subunit is used to identify the curb and lane lines in the road image, and calculate the vertical distance from the center of the manhole cover to the curb and the angle between the center of the manhole cover and the lane line; the distance calculation subunit is used to calculate the path distance between manhole covers at different time sequences; the combination subunit is used to combine the vertical distance from the center of the manhole cover to the curb, the angle between the center of the manhole cover and the lane line, the path distance, and the grid block ID to form a spatial topological feature value.

[0213] In one embodiment, the identity generation unit includes: a weight adjustment subunit, used to perform real-time quality analysis on the visual feature vector, spatial topological feature value, and physical tactile feature vector of the current frame, and dynamically adjust the weight coefficients of the visual feature vector, spatial topological feature value, and physical tactile feature vector according to the analysis results; and a conversion subunit, used to calculate a comprehensive feature value by combining the macroscopic attributes, visual feature vector, spatial topological feature value, and physical tactile feature vector and the corresponding weight coefficients using a weighted fusion strategy, and convert it into an identity identifier to obtain a manhole cover identity code.

[0214] In one embodiment, the weight adjustment subunit is used to reduce the corresponding weight and increase the weight of the remaining features when any one of the visual feature vector, spatial topological feature value, and physical tactile feature vector is unreliable due to interference from environmental factors.

[0215] In one embodiment, the comparison unit 307 includes:

[0216] The candidate search subunit is used to perform a retrieval operation in the database to find all known manhole cover files within a radius of several meters around the current manhole cover to obtain candidate matches. The similarity calculation subunit is used to calculate the similarity between the manhole cover's identity code and the fingerprint of the candidate match. When the similarity exceeds a set threshold and the relative topological location matches, the candidate match is a successfully matched object, confirming that the manhole cover is an existing record in the database. The addition subunit is used to mark the manhole cover as a new manhole cover, establish an identity code, and add the identity code to the database if multiple inspections fail to find a successfully matched object in the historical files but the same unidentified manhole cover exists.

[0217] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned dynamic inspection-based manhole cover coding system 300 and its various units can be referred to the corresponding descriptions in the aforementioned method embodiments. For the sake of convenience and brevity, these details will not be repeated here.

[0218] The aforementioned manhole cover coding system 300 based on dynamic inspection can be implemented as a computer program, which can, for example... Figure 8 It runs on the computer device shown.

[0219] Please see Figure 8 , Figure 8 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.

[0220] See Figure 8The 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.

[0221] 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 manhole cover coding method based on dynamic inspection.

[0222] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0223] 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 manhole cover coding method based on dynamic inspection.

[0224] This network interface 505 is used for network communication with other devices. Those skilled in the art will understand that... Figure 8 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.

[0225] The processor 502 is used to run the computer program 5032 stored in the memory to implement all the steps of the manhole cover coding method based on dynamic inspection.

[0226] 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.

[0227] 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.

[0228] 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 dynamic inspection-based manhole cover coding method.

[0229] 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.

[0230] 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.

[0231] In the several 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.

[0232] 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.

[0233] 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.

[0234] 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 manhole cover coding method based on dynamic inspection, characterized in that, include: Acquire road surface images captured by vehicle-mounted video capture equipment; The manhole cover target is extracted and located from the road surface image, and an orthophoto of the manhole cover is generated; Based on the analysis of the orthophoto of the manhole cover, the macroscopic attributes are analyzed, including administrative region, asset category, and filing time; Microscopic properties are analyzed based on the orthophoto of the manhole cover, wherein the microscopic properties include visual feature vectors and spatial topological feature values; The road surface vibration signal is acquired, and physical features are extracted from the road surface vibration signal using dynamic compensation and frequency domain analysis methods to obtain a physical tactile feature vector. Based on environmental changes, a manhole cover identification code is generated by combining the aforementioned macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors. The manhole cover identification code is compared with historical records in the database to confirm the manhole cover's identity or register it as a new asset.

2. The manhole cover coding method based on dynamic inspection according to claim 1, characterized in that, The process of acquiring road surface vibration signals and extracting physical features from these signals using dynamic compensation and frequency domain analysis methods to obtain physical tactile feature vectors includes: The road vibration signal is acquired, the deviation caused by the vehicle's shock absorption system is removed, and the amplitude is adjusted according to the vehicle speed to obtain the net impact signal. Dynamic gain compensation is then performed to obtain the processed signal. The inherent resonant frequency, spectral centroid, and energy decay rate are extracted from the processed signal to obtain physical characteristics characterizing the material properties of the manhole cover. The feature binning quantization method is used to map the values ​​of continuous physical features to discrete bucket IDs to obtain physical tactile feature vectors.

3. The manhole cover coding method based on dynamic inspection according to claim 2, characterized in that, The process of acquiring road vibration signals, removing deviations caused by the vehicle's shock absorption system, adjusting the amplitude according to vehicle speed to obtain a net impact signal, and performing dynamic gain compensation to obtain a processed signal includes: The road vibration signal is acquired, the mean value of the road background noise in the few seconds before the manhole cover is run over by the vehicle is calculated, and the deviation caused by the vehicle's shock absorption system is removed by the baseline elimination method to obtain the net impact signal. The amplitude of the net impact signal is adjusted to the equivalent value at the standard reference speed using a dynamic model based on the real-time GPS vehicle speed to obtain the processed signal.

4. The manhole cover coding method based on dynamic inspection according to claim 2, characterized in that, The extraction of the intrinsic resonant frequency, spectral centroid, and energy attenuation rate from the processed signal to obtain physical characteristics characterizing the material properties of the manhole cover includes: The processed signal is converted into a discrete power spectral density, and the sound features characterizing the inherent properties of the manhole cover material are analyzed from it to obtain the physical features characterizing the properties of the manhole cover material. This includes: finding the frequency point with the highest energy in the power spectrum as the inherent resonant frequency; calculating the centroid of the vibration energy distribution in the frequency domain to form the spectral centroid; and evaluating the rate of attenuation of the signal envelope after the impact peak to obtain the energy attenuation rate.

5. The manhole cover coding method based on dynamic inspection according to claim 2, characterized in that, The method of using feature bucketing quantization maps the values ​​of continuous physical features to discrete bucket IDs to obtain physical tactile feature vectors, including: A preset frequency tolerance step size is used to map the values ​​of continuous physical features to discrete bucket IDs to obtain physical tactile feature vectors.

6. The manhole cover coding method based on dynamic inspection according to claim 1, characterized in that, The method of analyzing microscopic attributes based on the orthophoto of the manhole cover includes visual feature vectors and spatial topological feature values, including: Microscopic key points are extracted from the surface of the manhole cover based on the orthophoto of the manhole cover to generate a visual feature vector; wherein, the microscopic key points include at least one of wear lines, rust spots, and text strokes; The curb and lane lines are identified in the road surface image, and the vertical distance from the center of the manhole cover to the curb and the angle between the center of the manhole cover and the lane line are calculated. Calculate the path distance between manhole covers at different time intervals; Get the grid block ID of the current location; The spatial topological feature value is formed by combining the vertical distance from the center of the manhole cover to the curb, the angle between the center of the manhole cover and the lane line, the path distance, and the grid block ID.

7. The manhole cover coding method based on dynamic inspection according to claim 1, characterized in that, The process of generating a manhole cover identification code based on environmental changes, combined with macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors, includes: The system performs real-time quality analysis on the visual feature vector, spatial topological feature value, and physical tactile feature vector of the current frame, and dynamically adjusts the weight coefficients of the visual feature vector, spatial topological feature value, and physical tactile feature vector based on the analysis results. A weighted fusion strategy is adopted to combine the macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors with corresponding weight coefficients to calculate a comprehensive feature value, which is then converted into an identity identifier to obtain the manhole cover identity code.

8. The manhole cover coding method based on dynamic inspection according to claim 7, characterized in that, The dynamic adjustment of the weight coefficients of the visual feature vector, spatial topological feature value, and physical tactile feature vector includes: When any of the visual feature vector, spatial topological feature value, or physical tactile feature vector becomes unreliable due to environmental factors, the corresponding weight is reduced, and the weights of the remaining features are increased.

9. The manhole cover coding method based on dynamic inspection according to claim 1, characterized in that, The step of comparing the manhole cover identification code with historical records in the database to confirm the manhole cover's identity or register it as a new asset includes: Perform a search operation in the database to find all known manhole cover files within a radius of several meters around the current manhole cover, in order to obtain candidate matches; Calculate the similarity between the manhole cover identification code and the fingerprint of the candidate matching item. When the similarity exceeds a set threshold and the relative topological location matches, the candidate matching item is a successfully matched object, confirming that the manhole cover is an existing record in the database. If multiple inspections fail to find a matching object in the historical archives, but the same unidentified manhole cover exists, mark the manhole cover as a newly added manhole cover, establish an identification code, and add the identification code to the database.

10. A manhole cover coding system based on dynamic inspection, characterized in that, include: The image acquisition unit is used to acquire road surface images captured by the vehicle-mounted video acquisition device; The image generation unit is used to extract and locate the manhole cover target from the road surface image and generate an orthophoto image of the manhole cover. The analysis unit is used to analyze macroscopic attributes based on the orthophoto of the manhole cover, wherein the macroscopic attributes include administrative region, asset category and filing time; The manhole cover differentiation unit is used to analyze microscopic attributes based on the orthophoto of the manhole cover, wherein the microscopic attributes include visual feature vectors and spatial topological feature values; The physical feature extraction unit is used to acquire road vibration signals and extract physical features from the road vibration signals using dynamic compensation and frequency domain analysis methods to obtain physical tactile feature vectors. The identity generation unit is used to generate a manhole cover identity code based on environmental changes and the macroscopic attributes, visual feature vectors, spatial topological feature values, and physical tactile feature vectors. The comparison unit is used to compare the manhole cover identification code with historical records in the database to confirm the manhole cover's identity or register it as a new asset.

Images