Dam inspection method and system based on augmented reality, medium and program product

Abstract

The invention provides a dam inspection method and system based on augmented reality, a medium and a program product, and relates to the technical field of reservoir dam safety monitoring, the method comprises the steps that a dam is divided into a plurality of physical areas according to objective risk data of each dam monitoring point, and each physical area corresponds to an objective risk value; obtaining a significance score corresponding to each physical region in the inspection process; calculating the attention value of each physical area according to the significance score and the real-time fixation point data of the inspection user; calculating a cognitive contrast index of each physical region in the polled region based on the objective risk value and the attention value, and further determining a target physical region; and according to the real-time position and orientation information of the inspection user, sequentially generating guide visual elements corresponding to each target physical area in the AR glasses to perform visual guidance on the inspection user. By implementing the method, the attention of the inspection personnel to the high-risk area which is not highlighted visually can be improved, and the probability of missing detection of the risk area is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of reservoir dam safety monitoring, and in particular to a dam inspection method and system based on augmented reality, a medium and a program product. BACKGROUND

[0002] With the continuous expansion and aging of infrastructure, the safety monitoring and inspection of large-scale water conservancy projects such as dams has become increasingly important. As a key link in preventing structural risks and ensuring public safety, dam inspection requires inspection personnel to conduct comprehensive and detailed detection of each part of the dam structure in order to timely discover and handle potential safety hazards. In order to improve the efficiency and accuracy of inspection, more and more dam management units have introduced digital and intelligent means in the inspection process, such as using high-precision sensors to collect real-time data on key parts of the dam, and using wearable devices to assist inspection personnel in obtaining on-site information, thereby improving the scientific nature and informatization level of dam safety management.

[0003] In related technologies, various types of sensors can be deployed at key parts of the dam to collect real-time monitoring data reflecting the health status of the dam structure, such as displacement, seepage pressure, crack width, etc. Inspection personnel can use tablets or AR (Augmented Reality) devices to superimpose monitoring data in the form of intuitive charts, color blocks, alarm markers, etc. on the structure diagram or three-dimensional model of the dam. During the inspection process, the system highlights or alarms part of the data according to the preset risk threshold, helping the inspector to identify areas that need to be paid attention to. In this way, related technologies can help inspection personnel quickly locate potential risk points, improve inspection efficiency, and reduce the burden of manual analysis.

[0004] However, related technologies mainly rely on direct visualization of sensor data and present relevant information to the inspector through uniform visual markers or alarm prompts, and the inspector often allocates attention according to the prominence of visual stimuli. In this way, in information-intensive or overly marked scenarios, some physical risk areas may not be prominent enough visually, or may be obscured by a large amount of other highlighted information, causing the inspector's attention to not evenly cover all risk points. In this case, there may be a deviation between the actual risk distribution of the physical risk area and the actual visual attention area of the inspector, resulting in the risk of missing key risk points. SUMMARY

[0005] The present application provides a dam inspection method and system based on augmented reality, a medium and a program product, which are used to address the problem that the actual physical risk area may not match the area where the user's visual attention is located during the dam inspection process, resulting in the risk of missing key risk areas.

[0006] In a first aspect, the application provides a dam inspection method based on augmented reality, applied to a dam inspection system, the method comprising: obtaining objective risk data of each dam monitoring point, the objective risk data being sensor values reflecting the actual risk level of the dam monitoring point; dividing the dam into a plurality of physical regions according to the position distribution corresponding to the objective risk data, each physical region corresponding to an objective risk value, the objective risk value being calculated according to the objective risk data; during the inspection process of the inspection user wearing AR glasses, obtaining the saliency score corresponding to each physical region by a computer vision saliency detection algorithm to obtain visual saliency data, the saliency score being a probability reflecting the attention of each physical region by the human eye; calculating the attention value of each physical region in the inspected region of the inspection user according to the visual saliency data and the real-time gaze point data of the inspection user, the real-time gaze point data being obtained by capturing the movement data of the user's eyeball through the AR glasses; calculating the cognitive contrast index of each physical region in the inspected region of the user based on the objective risk value and the attention value, the cognitive contrast index being positively correlated with the difference between the objective risk value and the attention value; determining one or more physical regions with the cognitive contrast index greater than a preset threshold as target physical regions; generating a guide visual element corresponding to each target physical region in the AR glasses according to the real-time position and orientation information of the inspection user to visually guide the inspection user, the guide visual element being a virtual graphic with direction, position and dynamic effect attributes.

[0007] Through the above embodiments, the dam inspection system combines the objective risk data of the dam with the actual visual attention behavior of the user, captures the gaze point of the user in real time through the AR glasses, and uses the visual saliency detection algorithm to comprehensively analyze the risk and attention degree of the physical region, finally dynamically generates a guide visual element to accurately guide the inspector to pay attention to the high-risk but easily overlooked region. Specifically, by mapping the risk data collected by the sensor to each physical region of the dam three-dimensional model, a regional risk distribution is formed; the real-time gaze point of the inspector is tracked by the AR glasses, combined with the visual saliency score, and the attention value of each region is quantified; further, by calculating the difference between the risk value and the attention value, the cognitive contrast index is obtained, and the target region that needs to be paid attention to is selected. Finally, the system provides dynamic and directional visual guidance for the inspector according to the position and orientation of the inspector. The above method improves the attention of the inspector to the high-risk but visually unobtrusive region, reduces the probability of missing the risk region, and substantially enhances the comprehensiveness and pertinence of the inspection.

[0008] In some embodiments, the step of dividing the dam into a plurality of physical regions according to the position distribution corresponding to the objective risk data specifically comprises: In the three-dimensional model of the dam, the spatial position coordinates of each dam monitoring point are matched with the objective risk data in position to generate a monitoring point distribution model with risk attributes; In the monitoring point distribution model, adjacent dam monitoring points with a difference in objective risk data within a preset change threshold are divided into the same physical region; According to the difference between the objective risk data and the standard risk data, the objective risk value corresponding to each physical region is calculated.

[0009] Through the above embodiments, the dam inspection system accurately matches the spatial position of the dam monitoring point with the risk data, and groups the monitoring points by difference in the three-dimensional model, so that the dam structure risk distribution not only has the continuity of spatial scale, but also dynamically reflects the risk difference, thereby realizing the scientific division of physical regions, improving the fine expression of risk distribution and the scientificity of regional division, and providing a spatial basis for subsequent risk visualization, key guidance and data analysis.

[0010] In some embodiments, the step of calculating the attention value of each physical region in the user-inspected region according to the visual saliency data and the real-time gaze point data of the inspection user specifically comprises: determining the physical region to which the real-time gaze point data of the inspection user belongs; In the inspection process, the cumulative gaze duration of the inspection user's gaze point in each physical region is counted; The cumulative gaze duration of each physical region is multiplied by the visual saliency score of the corresponding physical region of the cumulative gaze duration to obtain the attention value of each physical region in the inspected region.

[0011] Through the above embodiments, the dam inspection system combines the real-time gaze behavior of the inspection user with the visual saliency of the physical region, and obtains the attention value of each physical region by quantifying the product of the gaze duration and the saliency score, so that the system can truly reflect the actual attention degree of the inspector to each physical region during the inspection. Among them, the gaze duration reflects the actual behavior of the inspector, and the saliency score measures the visual attraction of the region, and the combination of the two can effectively avoid the attention deviation caused by relying only on visual stimulation.

[0012] In some embodiments, the step of calculating the attention value of each physical region in the user-inspected region according to the visual saliency data and the real-time gaze point data of the inspection user specifically comprises: Map the real-time gaze point data collected by the AR glasses worn by each inspection user to a unified spatial coordinate system, the unified spatial coordinate system being a three-dimensional reference coordinate system corresponding to a three-dimensional model of the dam; Aggregate the real-time gaze point data in the unified spatial coordinate system within a preset time window to obtain aggregated gaze point data, the aggregation being a merging and statistical calculation of the gaze point data of different inspection users in the same physical region; According to the weight coefficients corresponding to the task division of different inspection users, the aggregated gaze point data is weighted and calculated to obtain the attention values of each physical region.

[0013] Through the above embodiments, the dam inspection system can not only reflect the attention distribution of a single user, but also objectively integrate the overall attention of team inspection, thereby improving the comprehensiveness of risk monitoring and the fairness of data under team cooperation.

[0014] In some embodiments, the step of calculating the cognitive contrast index of each physical region in the user-inspected region based on the objective risk value and the attention value specifically includes: Obtain the spatial visibility parameter of each physical region, the spatial visibility parameter being an index reflecting whether the inspection user can directly view the physical region at the current position and orientation; Based on the spatial visibility parameter, dynamically correct the attention value of each physical region to obtain a corrected attention value; According to the objective risk value and the corrected attention value, calculate the cognitive contrast index of each physical region in the user-inspected region.

[0015] Through the above embodiments, in the process of calculating the cognitive contrast index based on the risk value and the attention value, the dam inspection system further introduces the spatial visibility parameter of the physical region and dynamically corrects the attention value, which can more truly and accurately reflect the actual inspection coverage of each physical region in the spatial environment, and reduce the risk of omission due to limited visibility.

[0016] In some embodiments, before the step of guiding the inspection user by sequentially generating in the AR glasses the guide visual elements corresponding to each of the target physical regions according to the real-time position and orientation information of the inspection user, the method further includes: Collect real-time position data and orientation data of the user, the real-time position data being the current position of the inspection user in the three-dimensional reference coordinate system corresponding to the three-dimensional model of the dam, and the orientation data being the line-of-sight direction vector of the inspection user; acquiring position data of each target physical region, the position data being spatial coordinates of the target physical region in the three-dimensional reference coordinate system; calculating a spatial distance between each target physical region and a current position of the inspection user, the spatial distance being an Euclidean distance between the current position of the user and the position data of the target physical region; calculating an included angle between each target physical region and the user orientation, the included angle being an included angle between the user orientation data and a connecting vector of the target physical region; calculating a priority score of each target physical region based on the spatial distance, the included angle of each target physical region, and an objective risk value of the target physical region, the priority score being a numerical value reflecting a comprehensive weight of the spatial distance, the included angle, and the risk value; determining a display order of the guiding visual elements corresponding to each target physical region in a descending order of the priority score of each target physical region.

[0017] Through the above embodiments, the dam inspection system intelligently fuses the precise calculation of the physical space and the risk weight, so that the areas with high risk and easy to ignore can be given priority to obtain the attention of the inspectors, and the spatial optimal ordering of the inspection route and the attention points is automatically realized, which greatly improves the inspection efficiency and reduces the probability of occurrence of high-risk blind spots.

[0018] In some embodiments, after the step of visually guiding the inspection user by sequentially generating the guiding visual elements corresponding to each target physical region in the AR glasses according to the real-time position and orientation information of the inspection user, the method further comprises: counting a number of times of being ignored by multiple inspection users for each target physical region, the number of times of being ignored being a cumulative number of times of not effectively paying attention to the target physical region by the inspection users within a preset time window; determining a target physical region with a number of times of being ignored exceeding a preset threshold number of times as a marked region; allocating an inspection task of the marked region to other available inspection users, the available inspection users being inspection users who are not currently allocated to the inspection task of the marked region; generating the guiding visual elements corresponding to the marked region in the AR glasses for the available inspection users.

[0019] Through the above embodiments, the dam inspection system counts the cumulative number of times of being ignored for each target region, and marks the target regions exceeding the threshold number of times as regions that need to be highlighted. Subsequently, the system automatically allocates the inspection tasks of these marked regions to available inspection users who have not been allocated the same type of task, and generates the corresponding guiding visual elements in the AR glasses of the available inspection users, thereby realizing the dynamic closed-loop adjustment of the inspection tasks, timely making up for the subjective omissions in the inspection, improving the coverage and integrity of the inspection work, and ensuring that the key risk regions will not be missed due to human subjectivity or divisional omissions.

[0020] In a second aspect, the present application provides a dam inspection system, comprising: one or more processors and a memory; The memory is coupled to the one or more processors, and the memory is configured to store computer program codes including computer instructions, and the one or more processors are configured to invoke the computer instructions so that the dam inspection system can implement the augmented reality-based dam inspection method provided in the above embodiments, which will not be repeated here.

[0021] In a third aspect, the present application provides a computer-readable storage medium comprising instructions that, when executed on a dam inspection system, cause the dam inspection system to implement the augmented reality-based dam inspection method provided in the above embodiments, which will not be repeated here.

[0022] In a fourth aspect, the present application provides a computer program product that, when executed on a dam inspection system, causes the dam inspection system to implement the augmented reality-based dam inspection method provided in the above embodiments, which will not be repeated here.

[0023] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. The objective risk data of each physical area of the dam is combined with the real-time visual attention behavior collected by the inspection user through the AR glasses, and the cognitive contrast index between the objective risk value and the inspection attention value is calculated to effectively identify the high-risk but easily ignored areas. This "risk-attention fusion" breaks through the limitations of traditional methods that rely only on sensor data or artificial subjective experience, enabling the system to dynamically and intelligently discover potential attention blind spots in the inspection, thereby greatly improving the comprehensiveness and pertinence of the inspection.

[0024] 2. Through spatial coordinate mapping of the inspection user's gaze point data, visibility parameter correction, and weighted aggregation of multiple inspection personnel data, spatial and team multi-dimensional attention modeling is achieved. Combined with the actual position, orientation, and task division of the user, the system can dynamically adjust the attention evaluation of each area and generate spatial-optimal and risk-oriented visual guidance elements for the inspection personnel according to the priority ranking. This collaborative mechanism not only maximizes the overall effect of the team but also improves the spatial coverage efficiency and intelligent division level of the inspection process, ensuring that high-risk areas are given priority attention.

[0025] 3. Through the statistics of the number of times the target area is ignored in the inspection process and the dynamic task reassignment, an intelligent leak-filling inspection closed-loop mechanism is constructed. Once some high-risk areas are ignored by multiple inspectors within a certain time window, the system will automatically assign these areas to new available inspectors and generate exclusive guiding visual elements in their AR glasses. This dynamic leak-filling mechanism significantly improves the integrity of the inspection work and the coverage of risk areas, effectively preventing important hidden dangers from being missed due to human subjectivity, information noise, or uneven task distribution, and achieving the goal of no dead angle and high reliability in dam inspection. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 is a flowchart of a dam inspection method based on augmented reality in an embodiment of the present application; Figure 2 is another flowchart of a dam inspection method based on augmented reality in an embodiment of the present application; Figure 3 is a schematic diagram of an entity device structure of a dam inspection system in an embodiment of the present application. DETAILED DESCRIPTION

[0027] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to be limiting on the present application. As used in the specification and the appended claims of the present application, the singular forms "a," "an," and "the" are intended to include both singular and plural forms, unless the context clearly indicates otherwise. It will be further understood that the term "and / or" used in the present application means any or all possible combinations of one or more of the listed items.

[0028] Hereinafter, the terms "first" and "second" are only for the purpose of description, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "multiple" is two or more, unless otherwise specified.

[0029] For ease of understanding, the method provided by the present embodiment is described in the flow. Please refer to Figure 1 is a flowchart of the method to be determined in an embodiment of the present application.

[0030] S101, objective risk data of each dam monitoring point is acquired.

[0031] The dam monitoring points refer to specific point positions arranged at key parts of the dam structure (such as the dam body, dam foundation, impervious body, and slope) for collecting safety-related data, for real-time or periodic monitoring of the operating state of the dam; the objective risk data refer to quantitative data that can reflect the actual risk level of the dam monitoring points, such as displacement, seepage pressure value, stress and strain value, crack width, water level change, and the like, which are directly collected by sensors.

[0032] Specifically, the dam inspection system establishes a communication connection with the sensors arranged at each monitoring point of the dam through a preset data interface, and the communication mode can include wired transmission (such as optical fiber, cable) or wireless transmission (such as LoRa, NB-IoT, 5G, etc.). The system actively acquires or passively receives the values sent by the sensors according to the set acquisition frequency (such as 1 time per second, 1 time every 5 minutes, etc.), and obtains the objective risk data of each dam monitoring point.

[0033] S102, dividing the dam into a plurality of physical regions according to the position distribution corresponding to the objective risk data, each physical region corresponding to an objective risk value.

[0034] The position distribution refers to the specific coordinate distribution of each dam monitoring point in the three-dimensional space of the dam, reflecting the spatial position relationship and layout characteristics of the monitoring points; the physical region refers to a number of continuous and similar risk attribute spatial regions obtained by dividing the dam according to the position distribution of the monitoring points and the characteristics of the objective risk data, and the risk conditions in each region are relatively consistent; and the objective risk value refers to a quantitative index calculated according to the objective risk data of all monitoring points in the physical region, for comprehensively reflecting the overall risk level of the physical region.

[0035] Specifically, the dam inspection system associates and matches the spatial position coordinates (such as X, Y, and Z coordinates under a three-dimensional coordinate system) of each monitoring point with the corresponding objective risk data, and establishes a monitoring point data set containing position and risk attributes. Then, based on the position distribution of the monitoring points, the region division rule (such as spatial clustering, adjacent merging, etc.) is used to classify the monitoring points that are adjacent in space and have small differences in objective risk data into the same physical region. For each physical region divided, the system calculates the objective risk value that can represent the risk level of the region by calculating the objective risk data of all monitoring points in the region (such as weighted average, maximum value extraction, comparison with standard value, etc.), thereby realizing the regionalization representation of the dam risk.

[0036] Optionally, the system imports a high-precision three-dimensional model of the dam, accurately maps the spatial coordinates of each monitoring point to the corresponding position in the model, and uses a density clustering algorithm (such as the DBSCAN algorithm) to cluster monitoring points that meet distance and data difference thresholds into a class based on the spatial distance and objective risk data difference of the monitoring points, forming a physical region. The objective risk data in each physical region is weighted (weights are set according to the importance of the monitoring point) to obtain the objective risk value of the region.

[0037] Optionally, the dam is segmented into preliminary large regions based on its structure (such as the dam crest section, dam body section, dam foundation section, spillway section, etc.). In each large region, the objective risk data difference of adjacent monitoring points is calculated, and adjacent monitoring points with a difference less than a preset change threshold are merged into a sub-region, i.e., a physical region. The average value of the difference between the objective risk data in each physical region and the preset standard risk data is calculated, and the average value is taken as the objective risk value of the region.

[0038] It can be understood that the physical region division can also be assisted by the structural characteristics of the dam (such as dam type, material zoning, stress characteristics, etc.), which is not limited here.

[0039] S103, during the inspection process of the inspection user wearing AR glasses, the visual saliency data is obtained by acquiring the saliency score corresponding to each physical region through a computer vision saliency detection algorithm.

[0040] The inspection user refers to a staff member wearing AR glasses responsible for performing dam inspection tasks, and conducts safety checks on each region of the dam through on-site observation and AR device assistance. The AR glasses refer to a wearable smart device with augmented reality functionality, integrating a camera, an eye tracking module, a display screen, and other components, capable of capturing on-site images, tracking user gaze, and superimposing virtual visual elements onto the real scene. The computer vision saliency detection algorithm refers to an algorithm for analyzing the visual appeal of each region in an image or video by extracting visual features such as color contrast, brightness difference, texture features, and edge information. The saliency score is a quantitative value obtained by the computer vision saliency detection algorithm, used to represent the probability of each physical region being noticed by the inspection user in terms of vision. The higher the score, the more likely the region is to be focused on by the human eye. The visual saliency data refers to a dataset containing all physical regions and their corresponding saliency scores.

[0041] Specifically, when the inspection user wears AR glasses for inspection, the camera of the AR glasses collects image data of the dam site in real time, and transmits the image data to the dam inspection system in real time. The system calls a computer vision saliency detection algorithm to process the collected real-time image, first identifies the region range corresponding to each physical region in the image, and then extracts the visual features (such as color saturation, brightness contrast, texture complexity, dynamic change, etc.) of these regions. The algorithm analyzes and calculates based on these visual features to generate a corresponding saliency score for each physical region. The higher the score, the more visually noticeable the region is to the inspection user. The system integrates the saliency scores of all physical regions to form visual saliency data.

[0042] Optionally, the AR glasses collect real-time video streams and send them to the dam inspection system frame by frame. The system pre-processes each frame of image (such as denoising, image correction, distortion correction); calls a saliency detection model based on deep learning (such as an improved ResNet-UNet model), and inputs the pre-processed image into the model; the model outputs a saliency heat map of the image, and the system extracts the average probability value of each region in the heat map as the saliency score according to the location of each region in the image, and aggregates to form visual saliency data.

[0043] S104, according to the visual saliency data and the real-time gaze point data of the inspection user, calculate the attention value of each physical region in the inspected region of the inspection user.

[0044] Among them, the real-time gaze point data refers to the quantitative data reflecting the eye movement trajectory of the inspection user captured by the eye tracking module of the AR glasses, including the coordinate position of the user's line of sight in three-dimensional space, the gaze time, etc.; the inspected region refers to the physical region that the inspection user has reached and observed during the inspection process, which is usually determined by the user's position trajectory and gaze range; the attention value refers to an index for quantifying the actual attention degree of the inspection user to each physical region, which is calculated by integrating the visual saliency data and the real-time gaze point data.

[0045] Specifically, the dam inspection system obtains the eye movement data of the user in real time through the eye tracking module of the AR glasses, and analyzes to obtain the coordinates of the real-time gaze point in three-dimensional space. Then the gaze point coordinates are mapped to the physical region division result of the dam three-dimensional model to determine the physical region to which each gaze point belongs. Combined with the saliency score of the physical region in the visual saliency data, the system fuses and calculates the gaze behavior (such as gaze duration, gaze frequency) and the saliency score through a preset algorithm. For the inspected region, the system accumulates the effective gaze data of each physical region, and finally generates an attention value that can reflect the actual attention degree of the user.

[0046] Optionally, the system collects user eye movement data in real time through the eye tracking module of the AR glasses, analyzes the two-dimensional coordinates of the fixation point in each frame of image, and maps them into the three-dimensional coordinate system of the dam; determines the physical region to which each real-time fixation point belongs, and counts the cumulative fixation duration of the region in the inspection process; multiplies the cumulative fixation duration by the visual saliency score of the physical region, and the product is the attention value of the physical region.

[0047] S105, calculate the cognitive contrast index of each physical region in the user-inspected region based on the objective risk value and the attention value.

[0048] Specifically, the dam inspection system retrieves the objective risk value and the corresponding attention value of each physical region in the inspected region. Then a preset calculation model is introduced, which needs to consider the difference between the two (e.g. the larger the difference, the higher the index). Before calculation, the system may correct the attention value (e.g. adjust the low attention value caused by visual obstruction by combining the spatial visibility parameter). Finally, the cognitive contrast index of each physical region is calculated through the model. The higher the index, the greater the deviation between the actual risk level of the region and the user's attention, and the more visual guidance is needed in the future.

[0049] Optionally, the system obtains the objective risk value (denoted as R) and the attention value (denoted as A) of each physical region in the inspected region; introduces a spatial visibility parameter K (range 0-1, 1 represents full visibility), corrects the attention value: A' = A x K; uses the formula to calculate the cognitive contrast index: I = (R - A') x ω, where ω is the weight coefficient (set according to the dam risk level), and the index I is positively correlated with (R - A').

[0050] Optionally, the system standardizes the objective risk value R and the corrected attention value A' (e.g. maps to the range of 0-100); calculates the absolute difference D = |R - A'|, and combines the risk level coefficient α (α is larger for high-risk regions); uses the formula to calculate the cognitive contrast index: I = D x α, to ensure that the difference value of high-risk regions has a greater impact on the index, and to achieve a positive correlation between the index and the difference value.

[0051] It can be understood that the relationship between the two can also be fitted by a machine learning model (such as a trained regression model) to obtain the index, which is not limited here.

[0052] S106, determine one or more physical regions with a cognitive contrast index greater than a preset threshold as target physical regions.

[0053] Specifically, the dam inspection system obtains the cognitive contrast index of each physical region in the inspected area, and then compares each index with a preset threshold. For the physical region whose index is greater than the threshold, the system marks it as a target physical region and records its position, objective risk value and other key information.

[0054] It should be noted that the preset threshold refers to the critical value preset by the system to determine whether the cognitive contrast index is significant. The setting of the preset threshold needs to consider factors such as the safety level of the dam (e.g. the threshold of important dams is lower and more sensitive), historical missed detection data (e.g. the threshold of once-missed detection area is lowered), etc. Here, no limitation is made.

[0055] S107, generating a guide visual element corresponding to each target physical region in the AR glasses according to the real-time position and orientation information of the inspection user to visually guide the inspection user.

[0056] Among them, the real-time position refers to the current spatial coordinates of the inspection user in the dam three-dimensional model reference coordinate system; the orientation information refers to the vector data of the user's line of sight direction, reflecting the user's current gaze direction; the guide visual element refers to the virtual graphics generated by the system in the AR glasses, which has the attributes of direction (pointing to the target region), position (superimposed on the corresponding position of the real scene) and dynamic effect (such as flashing, flowing), used to guide the user to focus on the target physical region.

[0057] Specifically, the dam inspection system obtains the real-time position coordinates and orientation vector of the user through the positioning module (such as GPS, SLAM) of the AR glasses, and retrieves the spatial coordinates of the target physical region. The system calculates the relative position (such as Euclidean distance, direction angle) of the user and each target region, and determines the display angle of the guide element according to the user's orientation (to ensure that the user can clearly observe). Then generate targeted guide visual elements: for example, use arrows to point to the direction of the target region, use highlight boxes to mark the position of the region in the real scene, and use flashing or flowing effects to enhance visual appeal. The display order of the elements is arranged according to the priority of the target region (e.g. the higher the risk, the closer the distance, the higher the priority), and finally superimposed and displayed in the AR glasses, realizing accurate visual guidance.

[0058] In the above embodiment, the dam inspection system combines the objective risk data of the dam with the actual visual attention behavior of the user, captures the user's gaze point in real time through AR glasses, and uses a visual saliency detection algorithm to comprehensively analyze the risk and attention of the physical area, and finally dynamically generates a guided visual element to accurately guide the inspector to pay attention to the high-risk but easily overlooked area. Specifically, by mapping the risk data collected by the sensor to each physical area of the dam three-dimensional model, a regional risk distribution is formed; the real-time gaze point of the inspector is tracked by the AR glasses, combined with the visual saliency score, and the attention value of each area is quantified; further, by calculating the difference between the risk value and the attention value, a cognitive contrast index is obtained, and the target area that needs to be paid attention to is screened out. Finally, the system provides dynamic and directional visual guidance to the inspector according to his position and orientation. The above method improves the attention of the inspector to the area with high risk but not visually prominent, reduces the probability of missing the risk area, and substantially enhances the comprehensiveness and pertinence of the inspection.

[0059] The method provided by the present embodiment is further described in more detail below. Please refer to Figure 2 , which is another flowchart of a dam inspection method based on augmented reality provided in the present embodiment.

[0060] S201, match the spatial position coordinates of each dam monitoring point with the objective risk data in the dam three-dimensional model to generate a monitoring point distribution model with risk attributes.

[0061] Specifically, the dam inspection system loads a pre-constructed dam three-dimensional model, which includes the accurate spatial coordinate system of each structure of the dam. Then, the system imports the spatial position coordinates (such as obtained by GPS positioning or manual calibration) and the corresponding objective risk data (such as displacement, seepage pressure, etc. data collected by sensors) of each dam monitoring point. Through coordinate mapping algorithm, the spatial position coordinates of each monitoring point are accurately matched with the corresponding position in the dam three-dimensional model, ensuring that each monitoring point finds a unique spatial position in the three-dimensional model. Then, the system binds the objective risk data as attribute information to the corresponding monitoring point in the three-dimensional model, forming a monitoring point distribution model containing spatial position and risk data double attributes, which can intuitively show the distribution position of each monitoring point on the dam and its risk level.

[0062] S202, in the monitoring point distribution model, adjacent dam monitoring points with a difference value of objective risk data within a preset change threshold are divided into the same physical area.

[0063] Specifically, the dam inspection system first identifies all adjacent dam monitoring points in the monitoring point distribution model, and screens out adjacent monitoring point pairs with a spatial distance (such as Euclidean distance) less than a preset proximity threshold by calculating the spatial distance between the monitoring points. Then, the system calculates the objective risk data difference of each adjacent monitoring point pair, and compares the difference with a preset change threshold. For adjacent monitoring points with a difference less than or equal to the preset change threshold, the system classifies them into the same group; for monitoring points with a difference exceeding the threshold, the system classifies them into different groups. In this way, all monitoring points are gradually aggregated into multiple continuous groups, each group corresponding to a physical region, ensuring that the monitoring points in the same physical region have similar risk levels and are spatially connected.

[0064] Optionally, in the monitoring point distribution model, a spatial proximity threshold (such as 8 meters) and a risk data preset change threshold (such as a displacement difference of 5 mm) are set; starting from any monitoring point, search for adjacent monitoring points with a spatial distance less than 8 meters, calculate the risk data difference between the two; if the difference is less than 5 mm, merge the two into a temporary region, continue to search for adjacent points of all monitoring points in the temporary region and repeat the difference judgment, until no monitoring point meets the condition, forming a physical region.

[0065] Optionally, a clustering algorithm (such as K-means clustering) is used, taking the spatial coordinates and risk data of the monitoring points as clustering features; set the initial value of the number of clusters, calculate the spatial-risk comprehensive distance between the monitoring points through the algorithm, and cluster the monitoring points with closer distances into a class; verify the clustering result, if the risk data difference within the class exceeds the preset change threshold, split the cluster, and finally form physical regions that meet the conditions.

[0066] It can be understood that the structure segmentation of the dam (such as the dam crest and the dam body) can also be combined to assist in dividing the physical regions, which is not limited here.

[0067] S203、According to the difference between the objective risk data and the standard risk data, calculate the objective risk value corresponding to each physical region.

[0068] Specifically, the dam inspection system first obtains the objective risk data of all monitoring points in each physical region, and simultaneously retrieves the standard risk data (such as the design allowable value) of the corresponding monitoring indicators. For each monitoring point, the system calculates the difference between its objective risk data and the standard risk data (such as the actual displacement minus the allowable displacement), obtaining the risk deviation value of the individual monitoring point. Then, the system comprehensively calculates the risk deviation values of all monitoring points in the physical region, which can be calculated by weighted average (according to the importance of the monitoring points), taking the maximum value (highlighting the highest risk point), etc., to obtain an objective risk value that can represent the overall risk level of the physical region. The higher the objective risk value, the greater the deviation from the safety standard in the region, and the higher the risk.

[0069] S204, map the real-time gaze point data collected by the AR glasses worn by each inspection user into a unified spatial coordinate system.

[0070] Specifically, the dam inspection system obtains the device parameters (such as intrinsic parameters and extrinsic parameters) and initial positioning information of the AR glasses of each inspection user, and determines the conversion relationship (such as rotation matrix and translation vector) between the local coordinate system of the AR glasses and the three-dimensional reference coordinate system of the dam. When the AR glasses collect real-time gaze point data (represented by two-dimensional or three-dimensional coordinates in the local coordinate system of the glasses), the system calls a preset coordinate conversion algorithm to map the local coordinates to the unified spatial coordinate system (the three-dimensional reference coordinate system of the dam) through the conversion relationship. During the mapping process, the system dynamically calibrates the conversion parameters in combination with the real-time position data (such as AR glasses GPS or SLAM positioning results) of the user, to ensure that the gaze point data collected by different users and different devices are accurately aligned in the same coordinate system, and to eliminate the data deviation caused by the difference in spatial reference.

[0071] S205, aggregate the real-time gaze point data in the unified spatial coordinate system within a preset time window to obtain aggregated gaze point data.

[0072] Specifically, the dam inspection system sets preset time window parameters (such as start time and duration), and triggers data aggregation in the system according to the window period. In each time window, the system traverses all real-time gaze point data in the unified coordinate system, and determines the physical region to which each gaze point belongs through spatial coordinate matching. Then, the gaze point data in the same physical region is merged and counted: the total number of gazes (the cumulative number of gazes of different users on the region), the total gaze time (the total time of all users gazing on the region), the average gaze rate (total number of times / window duration), and other indicators are counted to form aggregated gaze point data. During the aggregation process, the system associates the timestamp of the data to ensure that only valid data within the current time window is included, and to avoid interference from cross-window data.

[0073] Optionally, the system sets a fixed time window (such as every 30 minutes), and marks the time range of each window (such as 9:00-9:30, 9:30-10:00); at the end of each window, all gaze point data in the time period in the unified coordinate system is traversed, and is grouped according to the physical region; the total number of gazes and the total gaze time are counted for each group of data, to generate aggregated gaze point data in the window and store it.

[0074] Optionally, a sliding time window mechanism (window length 30 minutes, sliding step 10 minutes) is adopted; the timestamps of the gaze point data are monitored in real time, and the data within the current sliding window are dynamically included; for the gaze point data of the same physical region within the window, the cumulative gaze duration ratio (gaze duration of the region / total duration of the window) and the gaze video rate are calculated to form the real-time updated aggregated gaze point data.

[0075] It can be understood that the size of the time window can also be dynamically adjusted in combination with the risk level of the physical region, and the window of a high-risk region is smaller to improve the timeliness of the data, which is not limited here.

[0076] S206, according to the weight coefficients corresponding to the task division of different inspection users, the aggregated gaze point data is weighted and calculated to obtain the attention value of each physical region.

[0077] Specifically, the dam inspection system calls a preset user task division table to clearly define the responsible region, professional field (such as seepage monitoring, structural crack) and skill level (such as senior / junior inspector) of each inspection user. According to the task division table, the system matches a corresponding weight coefficient for each user: for example, the weight coefficient of the user responsible for the special inspection of a certain physical region is set to 1.2; the overall weight coefficient of a senior inspector is higher than that of a junior inspector (such as 1.1 vs 0.9). Then, the system splits the aggregated gaze point data according to the user source, weights the corresponding data (such as gaze duration, frequency) of each user with the weight coefficient of the user, and then sums up the weighted data of the same physical region to obtain the attention value of the region. During the weighting calculation process, the system will ensure that the sum of the weight coefficients is normalized (such as the sum of the weight proportions of all users is 1) to avoid the influence of data magnitude deviation on the result.

[0078] Optionally, an association matrix of task division and physical region is established (such as user A associated with regions 1-3, user B associated with regions 4-6); the weight coefficient of the user's gaze data in the associated region is 1.1, and the weight coefficient of the user's gaze data in the non-associated region is 0.9; after weighting the aggregated data of all users in the same physical region according to the above coefficients, the weighted sum of the total gaze duration is calculated as the attention value of the region.

[0079] It can be understood that the weight coefficient can also be dynamically adjusted through the historical inspection accuracy, which is not limited here.

[0080] S207, obtaining the spatial visibility parameter of each physical region to dynamically correct the attention value of each physical region, and obtaining the corrected attention value.

[0081] Specifically, the dam inspection system performs spatial visibility analysis on each physical region based on the dam three-dimensional model and the real-time position / orientation of the inspection user: through a ray tracing algorithm, the system simulates the path of the user's line of sight from the current position to the center of the region, and determines whether there is an obstruction (such as dam structure, equipment, vegetation). According to the degree of obstruction, the spatial visibility parameter is calculated: when there is no obstruction, the parameter is 1; when there is partial obstruction (such as 30% of the line of sight is blocked), the parameter is 0.7; when there is complete obstruction, the parameter is 0. Then, the system dynamically corrects the initial attention value using the spatial visibility parameter: for regions with low visibility parameters (such as ≤0.3), the weight of the attention value is reduced (such as multiplied by 0.5) to avoid misjudgment of low attention value as "ignore" due to objective unobservability; for high visibility regions (such as ≥0.8), the weight is maintained or increased (such as multiplied by 1.1) to highlight the risk of subjective ignore. The corrected attention value is more in line with the actual observation conditions, providing a reliable basis for the calculation of the cognitive contrast index.

[0082] Optionally, the system loads the dam three-dimensional model, marks the spatial coordinates of all potential obstructions (such as dam corners, monitoring equipment, trees); real-time acquisition of user position and orientation, use ray tracing algorithm to calculate the visibility ratio (visible area / total area) of each physical region; take the visibility ratio as the spatial visibility parameter, and correct it according to the formula "corrected attention value = initial value × (visibility parameter + 0.5)" (such as 1.5 times for full visibility, 0.5 times for complete obstruction).

[0083] S208、According to the objective risk value and the corrected attention value, calculate the cognitive contrast index of each physical region in the user's inspected region.

[0084] The dam inspection system retrieves the objective risk value and the corresponding corrected attention value of each physical region in the inspected region, ensuring one-to-one correspondence of data. Then, the system introduces a pre-set cognitive contrast calculation model, which takes the difference between the two as the core input: if the objective risk value is significantly higher than the corrected attention value, it indicates that the region has high risk but insufficient attention, and the contrast index increases; if the two are close, the contrast index is low. Before calculation, the system may unify the magnitude of the objective risk value and the corrected attention value (such as standardizing them to the range of 0-100), and then calculate through the formula (such as "contrast index = (objective risk value - corrected attention value) × risk level weight"). Finally, the cognitive contrast index of each physical region is generated, providing quantitative results for subsequent threshold judgment.

[0085] Optionally, the system can also call the objective risk value (R) and the corrected attention value (A') of each physical area; 2. Standardize R and A' (map to 0-100); 3. Use the formula "cognitive contrast index = (R - A') x ω, where ω is the risk level weight (high-risk area ω=1.2, medium-risk ω=1.0)", if R≤A', the index is 0, and the final cognitive contrast index is obtained.

[0086] S209, collect real-time position data and orientation data of the user, and position data of each target physical area.

[0087] The dam inspection system collects data in real time through the AR glasses worn by the inspection user: real-time position data is obtained through the positioning module (such as GPS, Beidou, SLAM visual positioning) integrated in the AR glasses, and is mapped to the dam three-dimensional reference coordinate system through coordinate conversion; the orientation data is jointly collected by the gyroscope, accelerometer and eye tracking module of the AR glasses, and a three-dimensional vector (such as horizontal angle, pitch angle) of the user's line of sight direction is generated. At the same time, the system calls the preset position data (such as the three-dimensional coordinates of the center of the area) of each target physical area from the database, to ensure that the data format is consistent with the user's position data (belonging to the dam three-dimensional reference coordinate system).

[0088] S210, calculate the spatial distance between each target physical area and the current position of the inspection user, and the included angle between each target physical area and the user's orientation.

[0089] The dam inspection system extracts the current position coordinates (X1, Y1, Z1) of the user and the position coordinates (X2, Y2, Z2) of the target physical area, and substitutes them into the Euclidean distance formula to calculate the spatial distance, obtaining the straight-line distance between the user and each target area. Then, the system converts the user's orientation data into a unit vector V1 (line of sight direction), and converts the vector (X2-X1, Y2-Y1, Z2-Z1) from the user to the target area into a unit vector V2. The included angle is calculated by the vector dot product formula: cosθ=V1・V2, θ=arccos(V1・V2), obtaining the included angle (θ) between the target area and the user's orientation (the smaller the θ, the closer the target to the user's current line of sight direction).

[0090] Optionally, the system can also use the Manhattan distance in three-dimensional space (|X2-X1|+|Y2-Y1|+|Z2-Z1|) to approximate the spatial distance, reduce the calculation complexity; through the image frame collected by the AR glasses camera in real time, the position of the target area in the image is recognized, and the pixel deviation angle of the target area from the image center is calculated as an approximate value of the included angle; the distance and the included angle results are smoothed to reduce the influence of instantaneous jitter.

[0091] S211、based on the spatial distance, the included angle, and the objective risk value of each target physical region, a priority score of each target physical region is calculated.

[0092] The dam inspection system sets comprehensive weights for the spatial distance, the included angle, and the objective risk value (for example, the objective risk value accounts for 50%, the spatial distance accounts for 30%, and the included angle accounts for 20%), and the weights can be dynamically adjusted according to the inspection stage (for example, the distance weight is high in the early stage, and the risk weight is high in the later stage). Then, the indexes are standardized: the spatial distance is converted according to “1 - distance / maximum distance” (the closer the distance, the higher the score); the included angle is converted according to “1 - included angle / 180°” (the smaller the included angle, the higher the score); and the objective risk value is converted according to “risk value / maximum risk value” (the higher the risk, the higher the score). Finally, the weighted sum of the standardized indexes is calculated according to the weights: priority score = (risk standardized value × 50%) + (distance standardized value × 30%) + (included angle standardized value × 20%). The system sorts the target regions according to the scores from high to low to determine the display order of the guiding visual elements.

[0093] S212、according to the priority scores of each target physical region from high to low, the display order of the guiding visual elements corresponding to the target physical regions is determined in sequence.

[0094] The dam inspection system aggregates the priority scores of all target physical regions to establish a “target region - priority score” correspondence table. Then, the system sorts the target physical regions according to the priority scores from high to low, with the region having the highest score listed as the first display object, and the initial display order is formed in sequence. During the sorting process, the system dynamically checks whether there are regions with the same score, and if there are, the secondary rules of “higher objective risk value priority” and “closer distance priority” are used to further refine the sorting. At the same time, the system associates the real-time position and orientation changes of the inspection user, and if the user's movement significantly improves the priority of a region (such as a sudden decrease in distance), the display order is dynamically adjusted to ensure that the guiding elements always match the current inspection state of the user.

[0095] Optionally, the system extracts the priority scores of each target physical region, sorts them from high to low, and generates a sorted list; for regions with adjacent scores in the sorted list that differ by less than 5%, the objective risk values are compared (the higher one has priority) to perform secondary sorting; the sorting result is used as the display order of the guiding visual elements, and is stored in the AR rendering queue, with the element having the highest priority displayed at the top in the AR field of view.

[0096] Optionally, the system can also set a dynamic sorting trigger condition (such as user movement exceeding 5 meters or change in direction exceeding 30°); when the trigger condition is met, the priority scores of each target area are recalculated and sorted; the display order of the guide visual elements is updated according to the new sorting result, and the original low-priority elements are delayed or reduced in size, and the high-priority elements are highlighted.

[0097] It can be understood that the display order can also be adjusted in combination with the inspection task progress, and the target areas that have not been completed are preferentially displayed, which is not limited here.

[0098] S213, the number of times each target physical area is ignored by multiple inspection users is counted, and the target physical area whose number of times of ignoring exceeds a preset number of times threshold is determined as a marked area.

[0099] Specifically, the dam inspection system sets a number of times of ignoring statistical counter for each target physical area, and the initial value is 0. Within a preset time window, the system monitors the position and gaze data of the inspection user in real time: when the user enters the vicinity of the target area (such as a spatial distance ≤ 5 meters) but the cumulative gaze time < a preset effective time (such as 10 seconds), or does not trigger the "checked" interactive confirmation through the AR glasses, it is determined as "not effectively paying attention", and the number of times of ignoring the corresponding area counter is added by 1. After the time window ends, the system compares the number of times of ignoring of each target area with the preset number of times threshold, and if the number of times exceeds the threshold, the area is marked as a "marked area", and its position, risk value and other information are recorded, and preparation is made for secondary inspection allocation.

[0100] S214, the inspection task of the marked area is allocated to other available inspection users for visual guidance.

[0101] Specifically, the dam inspection system retrieves the task status of the current inspection user, and selects available inspection users: excluding users who have been allocated the task of the marked area, and preferentially selecting users who are in an idle state, are close to the marked area (such as ≤ 20 meters), or are professionally matched (such as being responsible for the inspection of similar areas). Then, the system sends a task allocation instruction to the selected available inspection users, including the position, risk level and inspection focus of the marked area. At the same time, the system generates a guide visual element corresponding to the marked area in the AR glasses of the available user according to the real-time position and direction of the available user: such as a dynamic arrow pointing to the direction of the area, a highlight frame marking the boundary of the area, and a risk prompt (such as "high osmotic pressure area, focus on inspection") is superimposed. After allocation, the system tracks the task status in real time until the user completes effective attention or confirms the inspection.

[0102] The dam inspection system of the embodiment of the application is applied to an electronic device, Figure 3 An architectural schematic diagram of an electronic device suitable for implementing the embodiment of the application is shown.

[0103] It should be noted that, Figure 3 The electronic device shown is only an example and should not limit the function and use range of the embodiments of the present application.

[0104] Those skilled in the art can understand that all or part of the steps of various methods in the above embodiments can be completed or controlled by relevant hardware through instructions (computer programs), which can be stored in a computer readable storage medium and loaded and executed by a processor. The electronic device of the present embodiment includes a storage medium and a processor, wherein the storage medium stores a plurality of instructions which can be loaded by the processor to execute any step of the method provided by the embodiments of the present application.

[0105] Specifically, the storage medium and the processor are directly or indirectly electrically connected to realize the transmission or interaction of data. For example, these elements can be electrically connected to each other through one or more signal lines. The storage medium stores computer execution instructions for realizing the data access control method, including at least one software function module stored in the storage medium in the form of software or firmware, and the processor executes various function applications and data processing by running the software program and module stored in the storage medium. The storage medium can be, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. Among them, the storage medium is used to store programs, and the processor executes the programs after receiving execution instructions.

[0106] Further, the software programs and modules in the storage medium can also include an operating system, which can include various software components and / or drivers for managing system tasks, such as memory management, storage device control, power management, and the like, and can communicate with various hardware or software components to provide an operating environment for other software components. The processor can be an integrated circuit chip with a processing capability. The processor can be a general purpose processor, including a central processing unit (CPU), a network processor (NP), and the like, which can implement or execute the methods, steps, and logic flow diagrams disclosed in the embodiments. The general purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0107] Due to the instructions stored in the storage medium, the steps in any method provided by the embodiments of the present application can be executed, and thus the beneficial effects of any method provided by the embodiments of the present application can be achieved. Details are described in the foregoing embodiments, and thus will not be described here.

[0108] The above describes only the preferred embodiments of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for dam inspection based on augmented reality, applied to a dam inspection system, characterized in that, The method comprises: acquiring objective risk data of each dam monitoring point, the objective risk data being sensor values reflecting actual risk levels of the dam monitoring points; dividing the dam into a plurality of physical regions according to position distribution corresponding to the objective risk data, each physical region corresponding to an objective risk value, the objective risk value being calculated according to the objective risk data; in a process of inspection by a user wearing AR glasses, acquiring a saliency score corresponding to each physical region by a computer vision saliency detection algorithm to obtain visual saliency data, the saliency score being a probability of each physical region being noticed by the human eye; calculating attention values of each physical region in an inspected region of the user according to the visual saliency data and real-time gaze point data of the user, the real-time gaze point data being obtained by capturing movement data of the user's eyeball by the AR glasses; calculating a cognitive contrast index of each physical region in the inspected region of the user based on the objective risk value and the attention value, the cognitive contrast index being positively correlated with a difference between the objective risk value and the attention value; determining one or more physical regions with the cognitive contrast index greater than a preset threshold as target physical regions; generating a guide visual element corresponding to each target physical region in the AR glasses according to real-time position and orientation information of the user to visually guide the user, the guide visual element being a virtual figure with direction, position and dynamic effect attributes.

2. The method of claim 1, wherein, The step of dividing the dam into a plurality of physical regions according to position distribution corresponding to the objective risk data specifically comprises: matching spatial position coordinates of each dam monitoring point with the objective risk data in position in a three-dimensional model of the dam to generate a monitoring point distribution model with risk attributes; dividing adjacent dam monitoring points with a difference in the objective risk data within a preset change threshold range into the same physical region in the monitoring point distribution model; calculating an objective risk value corresponding to each physical region according to a difference between the objective risk data and standard risk data.

3. The method of claim 1, wherein, The step of calculating attention values of each physical region in an inspected region of the user according to the visual saliency data and real-time gaze point data of the user specifically comprises: determining a physical region to which the real-time gaze point data of the user belongs; in the inspection process, counting cumulative gaze time of the gaze point of the user in each physical region; multiplying the cumulative gaze time of each physical region with a visual saliency score of the corresponding physical region of the cumulative gaze time to obtain an attention value of each physical region in the inspected region.

4. The method of claim 1, wherein, The step of calculating attention values of each physical region in an inspected region of the user according to the visual saliency data and real-time gaze point data of the user specifically comprises: mapping real-time gaze point data collected by the AR glasses worn by each user to a unified spatial coordinate system, the unified spatial coordinate system being a three-dimensional reference coordinate system corresponding to the three-dimensional model of the dam; Aggregate the real-time gaze point data in the unified spatial coordinate system within a preset time window to obtain aggregated gaze point data, the aggregation being a merging and counting of gaze point data of different inspection users in the same physical region; According to the weight coefficients corresponding to the task division of different inspection users, weight distribution and weighted calculation are performed on the aggregated gaze point data to obtain attention values of each physical region.

5. The method of claim 1, wherein, The step of calculating the cognitive contrast index of each physical region in the user-inspected region based on the objective risk value and the attention value specifically includes: Obtain the spatial visibility parameter of each physical region, which is an index reflecting whether the inspection user can directly view the physical region at the current position and orientation; Based on the spatial visibility parameter, dynamically correct the attention value of each physical region to obtain a corrected attention value; According to the objective risk value and the corrected attention value, calculate the cognitive contrast index of each physical region in the user-inspected region.

6. The method of claim 1, wherein, Before the step of generating the guiding visual elements corresponding to each target physical region in the AR glasses according to the real-time position and orientation information of the inspection user to visually guide the inspection user, it further includes: Collect real-time position data and orientation data of the user, the real-time position data being the current position of the inspection user in the three-dimensional reference coordinate system corresponding to the dam three-dimensional model, and the orientation data being the line-of-sight direction vector of the inspection user; Obtain the position data of each target physical region, which is the spatial coordinates of the target physical region in the three-dimensional reference coordinate system; Calculate the spatial distance between each target physical region and the current position of the inspection user, which is the Euclidean distance between the current position of the user and the position data of the target physical region; Calculate the included angle between each target physical region and the user's orientation, which is the included angle between the user's orientation data and the connecting vector of the target physical region; Based on the spatial distance, the included angle of each target physical region, and the objective risk value of the target physical region, calculate the priority score of each target physical region, which is a numerical value reflecting the comprehensive weight of spatial distance, included angle, and risk value; According to the priority score of each target physical region from high to low, the display order of the guiding visual elements corresponding to each target physical region is determined in sequence.

7. The method of claim 1, wherein, After the step of generating the guiding visual elements corresponding to each target physical region in the AR glasses according to the real-time position and orientation information of the inspection user to visually guide the inspection user, it further includes: Statistically count the number of times each target physical region is ignored by multiple inspection users, the number of times being the cumulative number of times the target physical region is not effectively focused on by the inspection user within a preset time window; Determine the target physical region whose number of times of being ignored exceeds a preset number of times threshold as a marked region; Assign the inspection task of the marked region to other available inspection users, the available inspection users being inspection users who are not currently assigned to the task of the marked region; Generate the guiding visual elements corresponding to the marked region for the available inspection users in the AR glasses.

8. A dam inspection system characterized by, The dam inspection system comprises one or more processors and a memory; The memory is coupled to the one or more processors, and the memory is configured to store computer program code comprising computer instructions that, when invoked by the one or more processors, cause the dam inspection system to perform the method of any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instructions are run on a dam inspection system, they cause the dam inspection system to perform the method of any one of claims 1-7.

10. A computer program product, characterised in that, When the computer program product is run on a dam inspection system, it causes the dam inspection system to perform the method of any one of claims 1-7.

Images