A road construction site remote video monitoring method and system

Abstract

The application relates to the technical field of road construction remote monitoring, and particularly discloses a road construction site remote video monitoring method and system. The application comprises the following steps: constructing a construction site model; performing real-time video monitoring to obtain multiple real-time monitoring videos; obtaining multiple target appearance features; performing tracking identification on a remote monitoring target, selecting a target monitoring video, and marking a target monitoring area; creating a remote monitoring window, monitoring and displaying the target monitoring video, and synchronously displaying a region highlight of the construction site model. The application can perform tracking identification on the remote monitoring target in the multiple real-time monitoring videos, select the target monitoring video, monitor and display the target monitoring video, and synchronously display the region highlight of the construction site model, so that the most critical and most valuable monitoring video can be automatically displayed for remote managers, the target monitoring area can be synchronously displayed, and the remote monitoring efficiency is greatly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of road construction remote monitoring, and in particular relates to a method and system for remote video monitoring of a road construction site. Background Art

[0002] Remote monitoring of road construction is a technology that uses remote communication means to monitor and manage the construction site in real time during the road construction process. It can collect images, videos and various environmental parameters of the construction site in real time and transmit them to the designated monitoring center or the manager's mobile device through the network.

[0003] In existing technologies, road construction requires the deployment of a large number of surveillance cameras, which can obtain a large amount of different surveillance video data at the same time. However, in remote monitoring, it is difficult for remote managers to simply and quickly view the most critical and most valuable surveillance videos. Instead, they need to view each surveillance video one by one, which is time-consuming and labor-intensive, seriously affecting the efficiency of remote monitoring. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide a method and system for remote video monitoring of a road construction site, aiming to solve the problems raised in the background technology.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] A method for remote video monitoring of a road construction site, the method specifically comprising the following steps:

[0007] Obtain basic field data of the road construction site and build a construction site model;

[0008] Determining multiple video surveillance areas in the road construction site, performing real-time video surveillance in the multiple video surveillance areas, and acquiring multiple real-time surveillance videos;

[0009] Receiving remote monitoring requirements, performing target analysis, determining a remote monitoring target, and obtaining multiple target appearance features of the remote monitoring target;

[0010] Based on the plurality of target appearance features, tracking and identifying the remote monitoring target in the plurality of real-time monitoring videos, selecting the target monitoring video, and marking the target monitoring area;

[0011] A remote monitoring window is created to monitor and display the target monitoring video, and the construction site model is synchronously displayed with regional highlighting according to the target monitoring area.

[0012] As a further limitation of the technical solution of the embodiment of the present invention, obtaining basic field data of the road construction site and constructing the construction site model specifically include the following steps:

[0013] Obtain construction planning information for road construction;

[0014] Analyzing the construction planning information to determine the road construction site;

[0015] Acquiring basic field data of the road construction site;

[0016] The basic site data is modeled and a construction site model is constructed through cascade convolution.

[0017] As a further limitation of the technical solution of the embodiment of the present invention, determining multiple video surveillance areas in the road construction site, performing real-time video surveillance in the multiple video surveillance areas, and obtaining multiple real-time surveillance videos specifically include the following steps:

[0018] Identify multiple video surveillance areas within a road construction site and generate regional monitoring instructions;

[0019] According to the area monitoring instruction, real-time video monitoring is performed on the multiple video monitoring areas to obtain multiple real-time monitoring videos.

[0020] As a further limitation of the technical solution of the embodiment of the present invention, the receiving of the remote monitoring demand, performing target analysis, determining the remote monitoring target, and obtaining multiple target appearance features of the remote monitoring target specifically include the following steps:

[0021] Receive remote monitoring requests;

[0022] Performing target analysis on the remote monitoring requirements to determine the remote monitoring targets;

[0023] Acquiring multiple target backup images of the remote monitoring target from a preset target database;

[0024] Feature recognition is performed on the plurality of target backup images to obtain a plurality of target appearance features.

[0025] As a further limitation of the technical solution of the embodiment of the present invention, the tracking and identifying of the remote monitoring target in the multiple real-time monitoring videos based on the multiple target appearance features, selecting the target monitoring video, and marking the target monitoring area specifically include the following steps:

[0026] Generate tracking and identification instructions;

[0027] According to the tracking and identification instruction, based on the plurality of target appearance features, tracking and identifying the remote monitoring target in the plurality of real-time monitoring videos, and selecting the target monitoring video;

[0028] According to the target surveillance video, a corresponding target surveillance area is marked from the multiple video surveillance areas.

[0029] As a further limitation of the technical solution of the embodiment of the present invention, the creation of a remote monitoring window, monitoring and displaying the target monitoring video, and synchronously displaying the construction site model with highlighted areas according to the target monitoring area specifically include the following steps:

[0030] Create a remote monitoring window;

[0031] In the remote monitoring window, the target monitoring video is monitored and displayed;

[0032] Performing regional highlighting processing on the construction site model according to the target monitoring area to generate a regional highlight model;

[0033] In the remote monitoring window, the highlighted model of the region is synchronously displayed.

[0034] A road construction site remote video monitoring system, comprising a site model building unit, a real-time video monitoring unit, a monitoring target analysis unit, a target tracking and identification unit, and a remote monitoring display unit, wherein:

[0035] A site model building unit is used to obtain basic site data of the road construction site and build a construction site model;

[0036] A real-time video monitoring unit is used to determine multiple video monitoring areas in the road construction site, perform real-time video monitoring in the multiple video monitoring areas, and obtain multiple real-time monitoring videos;

[0037] A monitoring target analysis unit, configured to receive a remote monitoring request, perform target analysis, determine a remote monitoring target, and obtain a plurality of target appearance features of the remote monitoring target;

[0038] a target tracking and identification unit, configured to track and identify a remote monitoring target in the plurality of real-time monitoring videos based on the plurality of target appearance features, select a target monitoring video, and mark a target monitoring area;

[0039] The remote monitoring display unit is used to create a remote monitoring window, monitor and display the target monitoring video, and synchronously display the construction site model with regional highlighting according to the target monitoring area.

[0040] As a further limitation of the technical solution of the embodiment of the present invention, the field model building unit specifically includes:

[0041] Information acquisition module, used to obtain construction planning information of road construction;

[0042] An information analysis module, configured to analyze the construction planning information and determine the road construction site;

[0043] A data acquisition module, used to acquire basic field data of the road construction site;

[0044] The model building module is used to model the basic field data and build a construction site model through cascade convolution.

[0045] As a further limitation of the technical solution of the embodiment of the present invention, the monitoring target analysis unit specifically includes:

[0046] A demand receiving module is used to receive remote monitoring demands;

[0047] A target analysis module is used to perform target analysis on the remote monitoring requirements and determine the remote monitoring target;

[0048] An image acquisition module, configured to acquire a plurality of target backup images of the remote monitoring target from a preset target database;

[0049] The feature recognition module is used to perform feature recognition on the multiple target backup images to obtain multiple target appearance features.

[0050] As a further limitation of the technical solution of the embodiment of the present invention, the remote monitoring display unit specifically includes:

[0051] Window creation module, used to create remote monitoring window;

[0052] A monitoring display module, configured to display the target monitoring video in the remote monitoring window;

[0053] A highlight processing module, configured to perform regional highlight processing on the construction site model according to the target monitoring area to generate a regional highlight model;

[0054] A synchronous display module is used to synchronously display the regional highlight model in the remote monitoring window.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The embodiment of the present invention constructs a construction site model; performs real-time video monitoring, obtains multiple real-time monitoring videos; obtains multiple target appearance features; tracks and identifies remote monitoring targets, selects target monitoring videos, and marks target monitoring areas; creates a remote monitoring window, monitors and displays target monitoring videos, and simultaneously displays regional highlights on the construction site model. It is possible to track and identify remote monitoring targets in multiple real-time monitoring videos, select target monitoring videos, monitor and display target monitoring videos, and simultaneously display regional highlights on the construction site model, thereby automatically displaying the most critical and most valuable monitoring videos to remote managers, and simultaneously displaying the target monitoring areas, greatly improving the efficiency of remote monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention.

[0058] Figure 1 A flow chart of a method provided by an embodiment of the present invention is shown.

[0059] Figure 2 A flow chart of constructing a construction site model in the method provided by an embodiment of the present invention is shown.

[0060] Figure 3 A flow chart of obtaining multiple real-time monitoring videos in the method provided by an embodiment of the present invention is shown.

[0061] Figure 4 A flow chart of determining a remote monitoring target in a method provided by an embodiment of the present invention is shown.

[0062] Figure 5 A flow chart of target tracking and identification in the method provided by an embodiment of the present invention is shown.

[0063] Figure 6 A flowchart of target remote monitoring display in the method provided by an embodiment of the present invention is shown.

[0064] Figure 7 The application architecture diagram of the system provided by the embodiment of the present invention is shown.

[0065] Figure 8 The structure block diagram of the field model construction unit in the system provided by the embodiment of the present invention is shown.

[0066] Figure 9 It shows a structural block diagram of a monitoring target analysis unit in a system provided by an embodiment of the present invention.

[0067] Figure 10A structural block diagram of a remote monitoring display unit in a system provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0068] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0069] It can be understood that in the prior art, a large number of monitoring cameras need to be deployed for road construction, and a large amount of different monitoring video data can be obtained at the same time. However, in remote monitoring, it is difficult for remote management personnel to simply and quickly view the most critical and most valuable monitoring video, and it is necessary to view each monitoring video one by one, which is time-consuming and laborious and seriously affects the efficiency of remote monitoring.

[0070] To solve the above problems, an embodiment of the present application acquires basic site data of a road construction site, constructs a construction site model, determines a plurality of video monitoring areas in the road construction site, performs real-time video monitoring in the plurality of video monitoring areas, acquires a plurality of real-time monitoring videos, receives a remote monitoring requirement, performs target analysis, determines a remote monitoring target, and acquires a plurality of target appearance features of the remote monitoring target. Based on the plurality of target appearance features, the remote monitoring target is tracked and identified in the plurality of real-time monitoring videos, a target monitoring video is selected, and a target monitoring area is marked. A remote monitoring window is created, the target monitoring video is monitored and displayed, and the construction site model is regionally and synchronously highlighted and displayed according to the target monitoring area. The remote monitoring target can be tracked and identified in the plurality of real-time monitoring videos, the target monitoring video is selected, the target monitoring video is monitored and displayed, and the construction site model is regionally and synchronously highlighted and displayed, so as to automatically display the most critical and most valuable monitoring video for the remote management personnel, and the target monitoring area can be synchronously displayed, thereby greatly improving the efficiency of remote monitoring.

[0071] Figure 1 A flowchart of a method provided by an embodiment of the present application is shown.

[0072] Specifically, a road construction site remote video monitoring method, the method specifically includes the following steps:

[0073] Step S101, acquiring basic site data of a road construction site, constructing a construction site model.

[0074] In an embodiment of the present invention, by obtaining construction planning information for road construction and analyzing the construction planning information, the road construction site is determined, and then basic field data such as the map and BIM data of the road construction site are obtained, and then three-dimensional data extraction and modeling rendering processing are performed on the basic field data to construct a construction site model of the road construction site.

[0075] Specifically, Figure 2 A flow chart of constructing a construction site model in the method provided by an embodiment of the present invention is shown.

[0076] In a preferred embodiment of the present invention, obtaining basic on-site data of a road construction site and constructing a construction site model specifically include the following steps:

[0077] Step S1011, obtaining construction planning information of road construction;

[0078] Step S1012, analyzing the construction planning information to determine the road construction site;

[0079] Step S1013, obtaining basic site data of the road construction site;

[0080] Step S1014: Modeling the basic site data and constructing a construction site model through cascade convolution.

[0081] Specifically, the basic site data is modeled and a construction site model is constructed through cascade convolution. The specific steps are as follows:

[0082] Receive basic field data and perform noise filtering and format standardization to generate standardized data; use a spatial registration algorithm to align the video image in the standardized data with the 3D laser point cloud coordinates to obtain structured data that is aligned in time and space;

[0083] Extract the machine motion trajectory, personnel focus heat map, and clipping displacement from the spatiotemporally aligned structured data. Based on a sliding time window mechanism, dynamically evaluate the importance of the machine motion trajectory, personnel focus heat map, and clipping displacement. Update the weight value distribution map at preset intervals to obtain a three-dimensional weight matrix.

[0084] The cascaded convolutional architecture processes the spatiotemporally aligned structured data and the three-dimensional weight matrix. The shallow network processes the infrared and visible light features in the spatiotemporally aligned structured data to capture subtle visual features, and the deep network extracts macroscopic motion patterns, ultimately obtaining the output feature tensor.

[0085] According to the light change of the construction site, combined with the output feature tensor, the fusion ratio of infrared and visible light features is dynamically adjusted, for the image quality degradation caused by extreme weather, the feature restoration compensation is carried out by using the generative adversarial network, so as to obtain the environment robustness enhanced feature map;

[0086] The graph neural network is used to construct the construction surface three-dimensional topology with timestamp, the environment robustness enhanced feature map and three-dimensional laser point cloud coordinates are cross-modal fused to obtain the four-dimensional space-time situation model;

[0087] The current four-dimensional space-time situation model is compared with the historical four-dimensional space-time situation model to generate a feature offset report; based on the confidence in the feature offset report, the incremental training data set is screened and constructed; the current four-dimensional space-time situation model is trained by using the incremental training data set, and the optimized four-dimensional space-time situation model is obtained, after a preset number of iterations, the final four-dimensional space-time situation model is obtained, and the final four-dimensional space-time situation model is used as the construction site model.

[0088] Further, the road construction site remote video monitoring method further comprises the following steps:

[0089] Step S102, determine a plurality of video monitoring areas in the road construction site, and perform real-time video monitoring in the plurality of video monitoring areas to obtain a plurality of real-time monitoring videos.

[0090] In the embodiment of the application, by analyzing the basic site data, a plurality of video monitoring areas in the road construction site are determined, and a regional monitoring instruction is generated, according to the regional monitoring instruction, real-time video monitoring is performed on the plurality of video monitoring areas, and a plurality of real-time monitoring videos corresponding to the plurality of video monitoring areas are obtained.

[0091] Specifically, Figure 3 A flowchart for obtaining a plurality of real-time monitoring videos in the method provided by the embodiment of the application is shown.

[0092] In the preferred embodiment provided by the application, the determination of the plurality of video monitoring areas in the road construction site, the real-time video monitoring in the plurality of video monitoring areas, and the obtaining of the plurality of real-time monitoring videos specifically comprises the following steps:

[0093] Step S1021, determine a plurality of video monitoring areas in the road construction site, and generate a regional monitoring instruction;

[0094] Step S1022, according to the regional monitoring instruction, real-time video monitoring is performed on the plurality of video monitoring areas, and a plurality of real-time monitoring videos are obtained.

[0095] Specifically, the plurality of video monitoring areas in the road construction site are determined, and the regional monitoring instruction is generated, and the specific steps are as follows:

[0096] Obtain historical construction accident reports and construction schedules from basic field data. Analyze high-frequency risk points in historical construction accident reports and combine them with the construction schedule to establish a dynamic benchmark instruction template that includes regional hazard levels and job type sensitivity.

[0097] By linking monitoring equipment with dynamic benchmark instruction templates and performing three-dimensional coordinate calibration of key areas, point cloud reconstruction is then used to eliminate visual blind spots caused by equipment occlusion, generating a monitoring area dataset. This dataset includes the actual number of workers in each area, heavy machinery activity trajectories, the frequency of illegal operations, and the intensity of operations.

[0098] High-risk scenario characteristics were identified by analyzing high-frequency risk points in historical construction accident reports. A temporal convolutional neural network was used to analyze the correlation between the frequency of illegal operations and the intensity of work in each area of ​​the monitoring area dataset. High-risk scenario characteristics were then integrated to generate a priority assessment list.

[0099] Meteorological warning data is obtained from basic field data. The meteorological warning data includes real-time wind speed monitoring values ​​and rainfall intensity forecast parameters. Based on the high-risk work areas marked in the priority assessment list, a hierarchical response strategy is implemented in combination with the meteorological warning data to generate a priority environmental adaptive map. The priority environmental adaptive map includes the monitoring accuracy of the construction work area, the infrared coverage radius parameter in a rainstorm environment, and environmental sensitivity parameters.

[0100] Based on the monitoring accuracy of the construction area and the infrared coverage radius parameters in a rainstorm environment, the effective monitoring area parameters of each area are extracted and synchronously associated with the time window of the construction schedule to generate a spatiotemporal coupling queue. The spatiotemporal coupling queue includes spatiotemporal correlation parameters and process sensitivity coefficients.

[0101] Based on the spatiotemporal correlation parameters, the process sensitivity coefficient and the high-risk scene characteristics are dynamically weighted and fused, and then combined with the environmental sensitive parameters to generate regional monitoring instructions.

[0102] Furthermore, the road construction site remote video monitoring method further includes the following steps:

[0103] Step S103: receiving a remote monitoring requirement, performing target analysis, determining a remote monitoring target, and obtaining a plurality of target appearance features of the remote monitoring target.

[0104] In an embodiment of the present invention, when a remote management personnel needs to perform remote monitoring and management, the remote monitoring requirements of the remote management personnel are received, and then a target analysis is performed on the remote monitoring requirements to determine the remote monitoring target, and then multiple target backup images of the remote monitoring target are matched and obtained from a preset target database, and then feature recognition is performed on the multiple target backup images to obtain multiple target appearance features.

[0105] It is understandable that the remote monitoring targets can be construction workers, asphalt transport vehicles, tire rollers, vibratory rollers, graders, etc.

[0106] It is understandable that the types of feature recognition include color, shape, texture, etc.

[0107] Specifically, Figure 4 A flow chart of determining a remote monitoring target in a method provided by an embodiment of the present invention is shown.

[0108] In a preferred embodiment of the present invention, the steps of receiving a remote monitoring requirement, performing target analysis, determining a remote monitoring target, and obtaining multiple target appearance features of the remote monitoring target specifically include the following steps:

[0109] Step S1031, receiving a remote monitoring request;

[0110] Step S1032: performing target analysis on the remote monitoring demand to determine a remote monitoring target;

[0111] Step S1033, obtaining multiple target backup images of the remote monitoring target from a preset target database;

[0112] Step S1034: performing feature recognition on the plurality of target backup images to obtain a plurality of target appearance features.

[0113] Specifically, feature recognition is performed on a plurality of target backup images to obtain a plurality of target appearance features. The specific steps are as follows:

[0114] Determine the feature dimension classification template through the preset target database, read all target backup images, and establish an analysis channel corresponding to the feature dimension classification template for each target backup image to generate a feature dimension initialization image;

[0115] Using a deep learning model to perform layer-by-layer feature analysis on the initialized image of each feature dimension, generating a quantitative evaluation value on each feature dimension, and obtaining an initial evaluation value table, wherein the initial evaluation value table includes multiple initial evaluation values;

[0116] Determine the monitoring target category based on monitoring requirements; perform correlation analysis on the feature dimension initialization images under the same feature dimension based on the monitoring target category, calculate the numerical fluctuation pattern of the initial evaluation value in each feature channel between different images, and assign weights based on the strength of stability to generate the feature stability confidence level;

[0117] Environmental parameters are obtained from the monitoring area dataset. These parameters include light intensity and airborne particulate matter concentration. These parameters are processed through dynamic feature mapping to obtain dynamic environmental compensation parameters.

[0118] In combination with the construction schedule, the value decay curve of the target backup images collected at different time periods is analyzed, and the time compensation factor is determined based on the value decay curve of the target backup images collected at different time periods;

[0119] Obtain a feature template that matches the current target backup image, and measure the Bhattacharyya distance distribution between the actual extracted features and the standard features item by item to obtain the feature benchmark comparison value;

[0120] For each target backup image, the initial evaluation value and the feature stability confidence are nonlinearly superimposed, and the nonlinear superposition result is dynamically adjusted using the dynamic environment compensation parameter to obtain a multidimensional fusion feature vector; all multidimensional fusion feature vectors are weighted geometrically averaged using the time compensation factor to obtain a normalized feature vector; the normalized feature vector is difference-compensated using the feature benchmark comparison value to obtain a compensated feature vector; the compensated feature vector is encoded into a standardized format to obtain a standard feature vector; and the target appearance features are obtained using the standard feature vector.

[0121] Furthermore, the road construction site remote video monitoring method further includes the following steps:

[0122] Step S104 , based on the multiple target appearance features, tracking and identifying the remote monitoring target in the multiple real-time monitoring videos, selecting the target monitoring video, and marking the target monitoring area.

[0123] In an embodiment of the present invention, a tracking and identification instruction is generated, and then according to the tracking and identification instruction, feature matching and tracking identification of the remote monitoring target are performed in multiple real-time monitoring videos based on multiple target appearance features. From the multiple real-time monitoring videos, a target monitoring video with the remote monitoring target is selected, and according to the target monitoring video, the corresponding target monitoring area is marked from multiple video monitoring areas.

[0124] Specifically, Figure 5 A flow chart of target tracking and identification in the method provided by an embodiment of the present invention is shown.

[0125] In the preferred embodiments provided by the present application, the tracking and identification of the remote monitoring target based on the target appearance features in the real-time monitoring videos, the selection of the target monitoring video, and the marking of the target monitoring area specifically include the following steps:

[0126] Step S1041, generating a tracking and identification instruction;

[0127] Step S1042, tracking and identifying the remote monitoring target based on the target appearance features in the real-time monitoring videos according to the tracking and identification instruction, and selecting the target monitoring video;

[0128] Step S1043, marking the corresponding target monitoring area from the video monitoring areas according to the target monitoring video.

[0129] Specifically, the tracking and identification of the remote monitoring target based on the target appearance features in the real-time monitoring videos according to the tracking and identification instruction, and the selection of the target monitoring video specifically include the following steps:

[0130] The environment parameter value of the target monitoring point is obtained every interval preset period, and the environment parameter value includes a dust concentration parameter and an illumination intensity value. The dust concentration parameter is subjected to Gaussian filtering processing to obtain a processed dust concentration parameter. The illumination intensity value is converted to obtain a standardized brightness coefficient. A mapping relationship table is obtained based on the processed dust concentration parameter and the standardized brightness coefficient.

[0131] The key frames of each real-time monitoring video stream are extracted to obtain a data vector of feature dimensions. A standard feature is obtained based on a feature dimension classification template. The data vector of feature dimensions and the standard feature are compared dimension by dimension, and a multi-dimensional feature matching degree score matrix is generated. In the feature matching degree score matrix, the matching degree score corresponds to the environment adjustment parameter value one by one.

[0132] The dynamic adjustment coefficient of each feature dimension under the current environment parameter value is calculated through the mapping relationship table to obtain a feature dynamic adjustment coefficient. The matching degree score is nonlinearly transformed using the feature dynamic adjustment coefficient to obtain an environment adaptive feature weight.

[0133] The delay data from collection to processing completion is obtained by measuring each real-time monitoring video stream. The delay coefficient is obtained through the delay data. The attenuation index is obtained through the area danger level. The delay compensation coefficient is obtained by applying exponential attenuation multiplication to the delay coefficient using the attenuation index.

[0134] The distance attenuation coefficient is obtained based on BIM data, monitoring equipment coordinates, and monitoring equipment lens parameters. The multi-dimensional feature matching score matrix is ​​used to nonlinearly superimpose the feature matching scores of each dimension with the corresponding environmental adaptive feature weights to obtain the feature matching score of environmental impact. The feature matching score of environmental impact is adjusted using the delay compensation coefficient to generate a processed matching score. The processed matching score is optimized using the distance attenuation coefficient to obtain a comprehensive matching score.

[0135] Traverse all comprehensive matching scores and select the real-time surveillance video with the highest comprehensive matching score as the target surveillance video.

[0136] Furthermore, the road construction site remote video monitoring method further includes the following steps:

[0137] Step S105: Create a remote monitoring window, monitor and display the target monitoring video, and synchronously display the construction site model with highlighted areas according to the target monitoring area.

[0138] In an embodiment of the present invention, a remote monitoring window is created and equally divided into a monitoring sub-window A and a monitoring sub-window B. In the monitoring sub-window A of the remote monitoring window, the target monitoring video is monitored and displayed, and according to the spatial position of the target monitoring area at the road construction site, the construction site model is regionally highlighted to generate a regional highlight model. Then, in the monitoring sub-window B of the remote monitoring window, the regional highlight model is synchronously displayed, so that remote management personnel can not only view the target monitoring video of the remote monitoring target, but also clearly understand the position of the remote monitoring target at the road construction site.

[0139] Specifically, Figure 6 A flowchart of target remote monitoring display in the method provided by an embodiment of the present invention is shown.

[0140] In a preferred embodiment of the present invention, the steps of creating a remote monitoring window, monitoring and displaying the target monitoring video, and synchronously displaying the construction site model with highlighted areas according to the target monitoring area specifically include the following steps:

[0141] Step S1051, creating a remote monitoring window;

[0142] Step S1052: displaying the target surveillance video in the remote monitoring window;

[0143] Step S1053: performing regional highlighting processing on the construction site model according to the target monitoring area to generate a regional highlight model;

[0144] Step S1054: synchronously displaying the highlighted model of the region in the remote monitoring window.

[0145] Further, Figure 7 The application architecture diagram of the system provided by the embodiment of the present invention is shown.

[0146] In another preferred embodiment of the present invention, a road construction site remote video monitoring system includes:

[0147] The site model building unit 101 is used to obtain basic site data of the road construction site and build a construction site model.

[0148] In an embodiment of the present invention, the site model building unit 101 obtains construction planning information of the road construction, determines the road construction site by analyzing the construction planning information, and then obtains basic site data such as the map and BIM data of the road construction site, and then performs three-dimensional data extraction and modeling rendering processing on the basic site data to construct a construction site model of the road construction site.

[0149] Specifically, Figure 8 FIG. 1 shows a structural block diagram of the field model building unit 101 in the system provided by an embodiment of the present invention.

[0150] In a preferred embodiment of the present invention, the site model building unit 101 specifically includes:

[0151] Information acquisition module 1011, used to obtain construction planning information of road construction;

[0152] An information analysis module 1012 is used to analyze the construction planning information and determine the road construction site;

[0153] A data acquisition module 1013 is used to acquire basic field data of the road construction site;

[0154] The model building module 1014 is used to perform modeling processing on the basic site data and build a construction site model through cascade convolution.

[0155] Furthermore, the road construction site remote video monitoring system also includes:

[0156] The real-time video monitoring unit 102 is used to determine multiple video monitoring areas in the road construction site, perform real-time video monitoring in the multiple video monitoring areas, and obtain multiple real-time monitoring videos.

[0157] In an embodiment of the present invention, the real-time video monitoring unit 102 determines multiple video monitoring areas in the road construction site by analyzing basic field data, and then generates regional monitoring instructions. According to the regional monitoring instructions, real-time video monitoring is performed on the multiple video monitoring areas to obtain real-time monitoring videos corresponding to the multiple video monitoring areas.

[0158] The monitoring target analysis unit 103 is used to receive a remote monitoring requirement, perform target analysis, determine a remote monitoring target, and obtain multiple target appearance features of the remote monitoring target.

[0159] In an embodiment of the present invention, when a remote administrator needs to perform remote monitoring and management, the monitoring target analysis unit 103 receives the remote monitoring requirements of the remote administrator, then performs target analysis on the remote monitoring requirements, determines the remote monitoring target, and then matches and obtains multiple target backup images of the remote monitoring target from a preset target database, and then performs feature recognition on the multiple target backup images to obtain multiple target appearance features.

[0160] Specifically, Figure 9 It shows a structural block diagram of the monitoring target analysis unit 103 in the system provided by an embodiment of the present invention.

[0161] In a preferred embodiment of the present invention, the monitoring target analysis unit 103 specifically includes:

[0162] A demand receiving module 1031 is used to receive remote monitoring demands;

[0163] The target analysis module 1032 is used to perform target analysis on the remote monitoring demand and determine the remote monitoring target;

[0164] The image acquisition module 1033 is used to acquire multiple target backup images of the remote monitoring target from a preset target database;

[0165] The feature recognition module 1034 is configured to perform feature recognition on the plurality of target backup images to obtain a plurality of target appearance features.

[0166] Furthermore, the road construction site remote video monitoring system also includes:

[0167] The target tracking and identification unit 104 is configured to track and identify a remote monitoring target in the multiple real-time monitoring videos based on the multiple target appearance features, select a target monitoring video, and mark a target monitoring area.

[0168] In an embodiment of the present invention, the target tracking and identification unit 104 generates a tracking and identification instruction, and then performs feature matching and tracking identification of the remote monitoring target in multiple real-time monitoring videos based on multiple target appearance features in accordance with the tracking and identification instruction. From the multiple real-time monitoring videos, a target monitoring video with the remote monitoring target is selected, and according to the target monitoring video, a corresponding target monitoring area is marked from multiple video monitoring areas.

[0169] The remote monitoring display unit 105 is used to create a remote monitoring window, monitor and display the target monitoring video, and synchronously display the construction site model with highlighted areas according to the target monitoring area.

[0170] In an embodiment of the present invention, the remote monitoring display unit 105 creates a remote monitoring window and equally divides the remote monitoring window into a monitoring sub-window A and a monitoring sub-window B. In the monitoring sub-window A of the remote monitoring window, the target monitoring video is monitored and displayed, and according to the spatial position of the target monitoring area at the road construction site, the construction site model is regionally highlighted to generate a regional highlight model. Then, in the monitoring sub-window B of the remote monitoring window, the regional highlight model is synchronously displayed, so that the remote management personnel can not only view the target monitoring video of the remote monitoring target, but also clearly understand the position of the remote monitoring target at the road construction site.

[0171] Specifically, Figure 10 FIG. 1 shows a structural block diagram of the remote monitoring display unit 105 in the system provided by an embodiment of the present invention.

[0172] Among them, in the preferred embodiment provided by the present invention, the remote monitoring display unit 105 specifically includes:

[0173] Window creation module 1051, used to create a remote monitoring window;

[0174] A monitoring display module 1052 is used to monitor and display the target monitoring video in the remote monitoring window;

[0175] A highlight processing module 1053 is configured to perform regional highlight processing on the construction site model according to the target monitoring area to generate a regional highlight model;

[0176] The synchronous display module 1054 is used to synchronously display the regional highlight model in the remote monitoring window.

[0177] It should be understood that, although the various steps in the flow chart of each embodiment of the present invention are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0178] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When executed, the program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0179] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0180] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0181] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for remote video monitoring of a road construction site, characterized in that: The method specifically comprises the following steps: Obtain basic field data of the road construction site and build a construction site model; Determining multiple video surveillance areas in the road construction site, performing real-time video surveillance in the multiple video surveillance areas, and acquiring multiple real-time surveillance videos; Receiving remote monitoring requirements, performing target analysis, determining a remote monitoring target, and obtaining multiple target appearance features of the remote monitoring target; Based on the plurality of target appearance features, tracking and identifying the remote monitoring target in the plurality of real-time monitoring videos, selecting the target monitoring video, and marking the target monitoring area; Creating a remote monitoring window to monitor and display the target monitoring video, and synchronously displaying the construction site model with highlighted areas according to the target monitoring area; The acquisition of basic on-site data of the road construction site and the construction of the construction site model specifically include the following steps: Obtain construction planning information for road construction; Analyzing the construction planning information to determine the road construction site; Acquiring basic field data of the road construction site; Modeling the basic field data and constructing a construction site model through cascade convolution; The basic site data is modeled and a construction site model is constructed through cascade convolution. The specific steps are as follows: Receive basic field data and perform noise filtering and format standardization to generate standardized data; use a spatial registration algorithm to align the video image in the standardized data with the 3D laser point cloud coordinates to obtain structured data that is aligned in time and space; Extract the machine motion trajectory, personnel focus heat map, and clipping displacement from the spatiotemporally aligned structured data. Based on a sliding time window mechanism, dynamically evaluate the importance of the machine motion trajectory, personnel focus heat map, and clipping displacement. Update the weight value distribution map at preset intervals to obtain a three-dimensional weight matrix. The cascaded convolutional architecture processes the spatiotemporally aligned structured data and the three-dimensional weight matrix. The shallow network processes the infrared and visible light features in the spatiotemporally aligned structured data to capture subtle visual features, and the deep network extracts macroscopic motion patterns, ultimately obtaining the output feature tensor. Based on the lighting changes at the construction site and the output feature tensor, the fusion ratio of infrared and visible light features is dynamically adjusted. To address image quality degradation caused by extreme weather, a generative adversarial network is used to restore and compensate for feature degradation, resulting in a feature map with enhanced environmental robustness. A graph neural network is used to construct a three-dimensional topological structure of the construction surface with a timestamp. The feature map with enhanced environmental robustness is cross-modally fused with the three-dimensional laser point cloud coordinates to construct a four-dimensional spatiotemporal situation model. Compare the current four-dimensional space-time situation model with the historical four-dimensional space-time situation model to generate a feature offset report; based on the confidence level in the feature offset report, screen and construct an incremental training data set; use the incremental training data set to train the current four-dimensional space-time situation model and obtain an optimized four-dimensional space-time situation model. After a preset number of iterations, the final four-dimensional space-time situation model is obtained and used as the construction site model.

2. The road construction site remote video monitoring method according to claim 1, characterized in that: Determining multiple video surveillance areas in the road construction site, performing real-time video surveillance in the multiple video surveillance areas, and obtaining multiple real-time surveillance videos specifically include the following steps: Identify multiple video surveillance areas within a road construction site and generate regional monitoring instructions; According to the area monitoring instruction, real-time video monitoring is performed on the multiple video monitoring areas to obtain multiple real-time monitoring videos.

3. The road construction site remote video monitoring method according to claim 2, characterized in that: Identify multiple video surveillance areas within a road construction site and generate regional monitoring instructions. The specific steps are as follows: Obtain historical construction accident reports and construction schedules from basic site data; Analyze high-frequency risk points in historical construction accident reports and combine them with the construction schedule to establish a dynamic baseline instruction template that includes regional hazard levels and job type sensitivity; By linking monitoring equipment with dynamic benchmark instruction templates and performing three-dimensional coordinate calibration of key areas, point cloud reconstruction is then used to eliminate visual blind spots caused by equipment occlusion, generating a monitoring area dataset. This dataset includes the actual number of workers in each area, heavy machinery activity trajectories, the frequency of illegal operations, and the intensity of operations. High-risk scenario characteristics were identified by analyzing high-frequency risk points in historical construction accident reports. A temporal convolutional neural network was used to analyze the correlation between the frequency of illegal operations and the intensity of work in each area of ​​the monitoring area dataset. High-risk scenario characteristics were then integrated to generate a priority assessment list. Meteorological warning data is obtained from basic field data. The meteorological warning data includes real-time wind speed monitoring values ​​and rainfall intensity forecast parameters. Based on the high-risk work areas marked in the priority assessment list, a hierarchical response strategy is implemented in combination with the meteorological warning data to generate a priority environmental adaptive map. The priority environmental adaptive map includes the monitoring accuracy of the construction work area, the infrared coverage radius parameter in a rainstorm environment, and environmental sensitivity parameters. Based on the monitoring accuracy of the construction area and the infrared coverage radius parameters in a rainstorm environment, the effective monitoring area parameters of each area are extracted and synchronously associated with the time window of the construction schedule to generate a spatiotemporal coupling queue. The spatiotemporal coupling queue includes spatiotemporal correlation parameters and process sensitivity coefficients. Based on the spatiotemporal correlation parameters, the process sensitivity coefficient and the high-risk scene characteristics are dynamically weighted and fused, and then combined with the environmental sensitive parameters to generate regional monitoring instructions.

4. The road construction site remote video monitoring method according to claim 3, characterized in that: The receiving of remote monitoring requirements, performing target analysis, determining a remote monitoring target, and obtaining multiple target appearance features of the remote monitoring target specifically include the following steps: Receive remote monitoring requests; Performing target analysis on the remote monitoring requirements to determine the remote monitoring targets; Acquiring multiple target backup images of the remote monitoring target from a preset target database; Feature recognition is performed on the plurality of target backup images to obtain a plurality of target appearance features.

5. The road construction site remote video monitoring method according to claim 4, characterized in that: Perform feature recognition on multiple target backup images to obtain multiple target appearance features. The specific steps are as follows: Determine the feature dimension classification template through the preset target database, read all target backup images, and establish an analysis channel corresponding to the feature dimension classification template for each target backup image to generate a feature dimension initialization image; Using a deep learning model to perform layer-by-layer feature analysis on the initialized image of each feature dimension, generating a quantitative evaluation value on each feature dimension, and obtaining an initial evaluation value table, wherein the initial evaluation value table includes multiple initial evaluation values; Determine the monitoring target category based on monitoring requirements; perform correlation analysis on the feature dimension initialization images under the same feature dimension based on the monitoring target category, calculate the numerical fluctuation pattern of the initial evaluation value in each feature channel between different images, and assign weights based on the strength of stability to generate the feature stability confidence level; Environmental parameters are obtained from the monitoring area dataset. These parameters include light intensity and airborne particulate matter concentration. These parameters are processed through dynamic feature mapping to obtain dynamic environmental compensation parameters. In combination with the construction schedule, the value decay curve of the target backup images collected at different time periods is analyzed, and the time compensation factor is determined based on the value decay curve of the target backup images collected at different time periods; Obtain a feature template that matches the current target backup image, and measure the Bhattacharyya distance distribution between the actual extracted features and the standard features item by item to obtain the feature benchmark comparison value; For each target backup image, the initial evaluation value and the feature stability confidence level are nonlinearly superimposed. The nonlinear superposition result is then dynamically adjusted using the dynamic environment compensation parameter to obtain a multidimensional fusion feature vector. The time compensation factor is used to perform weighted geometric averaging on all multidimensional fusion feature vectors to obtain a normalized feature vector. The normalized feature vector is compensated for difference using the feature reference comparison value to obtain a compensated feature vector; the compensated feature vector is encoded into a standardized format to obtain a standard feature vector; and the target appearance feature is obtained using the standard feature vector.

6. The road construction site remote video monitoring method according to claim 5, characterized in that: The method of tracking and identifying a remote monitoring target in a plurality of real-time monitoring videos based on the plurality of target appearance features, selecting a target monitoring video, and marking a target monitoring area specifically includes the following steps: Generate tracking and identification instructions; According to the tracking and identification instruction, based on the plurality of target appearance features, tracking and identifying the remote monitoring target in the plurality of real-time monitoring videos, and selecting the target monitoring video; According to the target surveillance video, a corresponding target surveillance area is marked from the multiple video surveillance areas.

7. The road construction site remote video monitoring method according to claim 6, characterized in that: According to the tracking and identification instruction, based on the multiple target appearance features, tracking and identifying the remote monitoring target in the multiple real-time monitoring videos is performed, and the target monitoring video is selected. The specific steps are as follows: The environmental parameter values ​​of the target monitoring point are obtained at each preset interval, and the environmental parameter values ​​include dust concentration parameters and light intensity values; the dust concentration parameters are Gaussian filtered to obtain processed dust concentration parameters; the light intensity values ​​are converted to obtain a standardized brightness coefficient; and a mapping relationship table is obtained based on the processed dust concentration parameters and the standardized brightness coefficient; Extract key frames from each real-time monitoring video stream to obtain data vectors of feature dimensions; obtain standard features based on feature dimension classification templates; Compare the data vector of the feature dimension with the standard feature dimension by dimension, and generate a multi-dimensional feature matching score matrix; Calculate the dynamic adjustment coefficient of each feature dimension under the current environmental parameter value through the mapping relationship table to obtain the feature dynamic adjustment coefficient; The matching score is nonlinearly transformed using the dynamic adjustment coefficient of the feature to obtain the environment-adaptive feature weight; By measuring the delay data from acquisition to processing of each real-time monitoring video stream, the delay coefficient is obtained from the delay data; the attenuation index is obtained based on the regional hazard level; Applying exponential decay multiplication to the delay coefficient using the decay exponent to obtain the delay compensation coefficient; The distance attenuation coefficient is obtained based on BIM data, monitoring equipment coordinates, and monitoring equipment lens parameters. The multi-dimensional feature matching score matrix is ​​used to nonlinearly superimpose the feature matching scores of each dimension with the corresponding environmental adaptive feature weights to obtain the feature matching score of environmental impact. The feature matching score of the environmental impact is adjusted using the delay compensation coefficient to generate a processed matching score; the processed matching score is optimized using the distance attenuation coefficient to obtain a comprehensive matching score; Traverse all comprehensive matching scores and select the real-time surveillance video with the highest comprehensive matching score as the target surveillance video.

8. The road construction site remote video monitoring method according to claim 7, characterized in that: The step of creating a remote monitoring window, monitoring and displaying the target monitoring video, and synchronously displaying the construction site model with highlighted areas according to the target monitoring area specifically includes the following steps: Create a remote monitoring window; In the remote monitoring window, the target monitoring video is monitored and displayed; Performing regional highlighting processing on the construction site model according to the target monitoring area to generate a regional highlight model; In the remote monitoring window, the highlighted model of the region is synchronously displayed.

9. A road construction site remote video monitoring system, characterized in that: The system applies the road construction site remote video monitoring method according to any one of claims 1 to 8, and the system includes a site model building unit, a real-time video monitoring unit, a monitoring target analysis unit, a target tracking and identification unit, and a remote monitoring display unit, wherein: A site model building unit is used to obtain basic site data of the road construction site and build a construction site model; A real-time video monitoring unit is used to determine multiple video monitoring areas in the road construction site, perform real-time video monitoring in the multiple video monitoring areas, and obtain multiple real-time monitoring videos; A monitoring target analysis unit, configured to receive a remote monitoring request, perform target analysis, determine a remote monitoring target, and obtain a plurality of target appearance features of the remote monitoring target; a target tracking and identification unit, configured to track and identify a remote monitoring target in the plurality of real-time monitoring videos based on the plurality of target appearance features, select a target monitoring video, and mark a target monitoring area; The remote monitoring display unit is used to create a remote monitoring window, monitor and display the target monitoring video, and synchronously display the construction site model with regional highlighting according to the target monitoring area.

Images