Real-time video analysis method and system based on intelligent image processing

Abstract

The application provides a real-time video analysis method and system based on intelligent image processing, wherein the method comprises: acquiring multi-view video data of a traffic monitoring area and motion trajectory data of a vehicle; using an optical flow compensation algorithm to correct inter-frame offset of the multi-view video data to generate corrected multi-view video data; extracting geometric feature points from the corrected multi-view video data and constructing a topological relationship graph based on the geometric feature points; optimizing the topological relationship graph based on the motion trajectory data to generate an optimized topological relationship graph; and generating a video analysis result based on the optimized topological relationship graph and the corrected multi-view video data. The application improves the accuracy and real-time performance of multi-view video analysis in a complex traffic scene.

Description

Technical Field

[0001] This application relates to the field of multi-view video real-time analysis and 3D scene reconstruction technology in the field of intelligent traffic monitoring, and in particular to a real-time video analysis method and system based on intelligent image processing. Background Technology

[0002] In the fields of urban intelligent traffic management and security monitoring, large-scale open scenarios require real-time and accurate analysis of multi-view video data to obtain dynamic traffic flow information, anomaly detection, and 3D scene reconstruction. These scenarios require video analysis methods to handle differences in viewpoints between multiple cameras, compensation for moving target displacement, and data consistency in complex environments, while also ensuring real-time performance and robustness.

[0003] Currently, a typical approach employs multi-view video stitching technology based on feature point matching. This method extracts stable feature points from each camera's video feed, calculates the homography matrix for viewpoint alignment, and then combines this with a moving object detection algorithm to separate dynamic foreground objects from the static background. Finally, it fuses multiple video streams to output a panoramic monitoring image. This method relies on high-precision feature matching and utilizes offline calibration parameters to optimize the stitching effect.

[0004] In dynamic scenes, this scheme is prone to stitching misalignment due to the displacement of moving targets. Feature point matching is significantly affected by changes in lighting and occlusion, and its reliance on offline calibration makes it difficult to adapt to real-time changes in camera pose. Furthermore, incomplete separation of moving targets from the background can cause ghosting, affecting the accuracy of subsequent analysis. Summary of the Invention

[0005] This application provides a real-time video analysis method and system based on intelligent image processing to solve the problems of low accuracy and poor real-time performance in multi-view video analysis under complex traffic scenarios in the prior art.

[0006] In a first aspect, this application provides a real-time video analysis method based on intelligent image processing, including:

[0007] Acquire multi-view video data and vehicle trajectory data of the traffic monitoring area;

[0008] Using an optical flow compensation algorithm, inter-frame offset correction is performed on the multi-view video data to generate corrected multi-view video data;

[0009] Geometric feature points are extracted from the corrected multi-view video data, and a topological relationship graph is constructed based on the geometric feature points.

[0010] The topology graph is optimized based on the motion trajectory data to generate an optimized topology graph;

[0011] Based on the optimized topology diagram and the corrected multi-view video data, video analysis results are generated.

[0012] Optionally, optimizing the topology graph based on the motion trajectory data to generate an optimized topology graph includes:

[0013] Extract the trajectory position sequence from the motion trajectory data;

[0014] Map each location point in the trajectory location sequence to the corresponding node in the topological relationship graph;

[0015] For any two adjacent nodes in the topological graph that are directly connected by a topological edge, if the angle deviation between the movement direction angle of the two adjacent nodes and the preset direction angle of the topological edge connecting the two adjacent nodes exceeds a preset angle threshold, then the constraint weight value of the topological edge is updated.

[0016] Based on the updated constraint weight values, the weight information of the topological edges in the topological graph is optimized to generate an optimized topological graph.

[0017] Optionally, updating the constraint weight value of the topological edge when the angular deviation between the movement direction angle value of the two adjacent nodes and the preset direction angle value of the topological edge connecting the two adjacent nodes exceeds a preset angle threshold includes:

[0018] Calculate the movement direction angle value of the two adjacent nodes, and use the movement direction angle value as the first direction value;

[0019] The angle value of the preset direction of the topological edge connecting the two adjacent nodes is used as the second direction value;

[0020] Calculate the absolute difference between the first direction value and the second direction value;

[0021] If the absolute difference exceeds the preset angle threshold, the constraint weight value of the topological edge connecting the two adjacent nodes is multiplied by the incremental coefficient to generate an updated constraint weight value.

[0022] Optionally, generating video analysis results based on the optimized topology graph and the corrected multi-view video data includes:

[0023] The geometric feature points in the corrected multi-view video data are matched with the nodes in the optimized topology graph.

[0024] If a match fails, the geometric feature point corresponding to the failed match is determined as a new geometric feature point, and the minimum Euclidean distance between the new geometric feature point and all existing nodes is calculated.

[0025] When the minimum Euclidean distance exceeds a preset distance threshold, a new node corresponding to the newly added geometric feature point is inserted into the topology graph, and the connection set is updated.

[0026] Based on the motion trajectory data, verify whether the connection path between the new node and all existing nodes is valid;

[0027] If the verification finds that at least one path of the motion trajectory is inconsistent with the preset connected path in the updated topology graph, it is determined that there is a conflict. Then the constraint weight values ​​of the relevant topology edges are adjusted to obtain the adjusted topology graph.

[0028] Based on the corrected multi-view video data, the adjusted topology diagram, and the motion trajectory data, video analysis results are generated.

[0029] Optionally, inserting the new node corresponding to the newly added geometric feature point into the topology graph and updating the connection set includes:

[0030] Obtain the set of neighboring nodes within a preset radius of the new node;

[0031] Create a new node in the topology graph, and set the position coordinates of the new node to the coordinates of the newly added geometric feature point;

[0032] Establish connections between the new node and each node in the set of neighboring nodes, and add the connections to the set of connections.

[0033] Optionally, constructing a topological relationship graph based on the geometric feature points includes:

[0034] Based on the spatial location of the geometric feature points, the geometric feature points corresponding to the same physical location are clustered into a set of location features;

[0035] Generate a set of connection relationships based on the spatial adjacency relationships between adjacent sets of location features;

[0036] A topological graph is constructed using the set of location features as nodes and the set of connection relationships as topological edges.

[0037] Optionally, the step of using an optical flow compensation algorithm to perform inter-frame offset correction on the multi-view video data to generate corrected multi-view video data includes:

[0038] Using an optical flow compensation algorithm, the inter-frame motion vector between adjacent video frames in the multi-view video data is calculated;

[0039] Filter the set of valid movement regions from the inter-frame motion vectors;

[0040] The offset of the effective mobile region set is recursively accumulated, and the inter-frame offset correction parameter is calculated based on the accumulation result;

[0041] Adjust the position coordinates of the video frame according to the inter-frame offset correction parameters;

[0042] Based on the adjusted video frames, the multi-view video data is corrected to generate corrected multi-view video data.

[0043] Secondly, this application provides a real-time video analysis system based on intelligent image processing, comprising:

[0044] The acquisition module is used to acquire multi-view video data and vehicle motion trajectory data of the traffic monitoring area;

[0045] The first generation module is used to perform inter-frame offset correction on the multi-view video data using an optical flow compensation algorithm, and generate corrected multi-view video data.

[0046] The construction module is used to extract geometric feature points from the corrected multi-view video data and construct a topological relationship graph based on the geometric feature points;

[0047] The second generation module is used to optimize the topology graph based on the motion trajectory data and generate an optimized topology graph.

[0048] The third generation module is used to generate video analysis results based on the optimized topology diagram and the corrected multi-view video data.

[0049] Thirdly, this application provides a computing device including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform a real-time video analysis method based on intelligent image processing as described in any of the first aspects.

[0050] Fourthly, this application provides a computer storage medium storing computer program instructions thereon, which, when executed by a processor, implement the real-time video analysis method based on intelligent image processing as described in any one of the first aspects.

[0051] This application provides a real-time video analysis method based on intelligent image processing. The method includes: acquiring multi-view video data of a traffic monitoring area and vehicle motion trajectory data; using an optical flow compensation algorithm to perform inter-frame offset correction on the multi-view video data to generate corrected multi-view video data; extracting geometric feature points from the corrected multi-view video data and constructing a topology graph based on the geometric feature points; optimizing the topology graph based on the motion trajectory data to generate an optimized topology graph; and generating video analysis results based on the optimized topology graph and the corrected multi-view video data.

[0052] The technical solution provided in this application has the following beneficial effects:

[0053] This application achieves synchronous acquisition of multi-source heterogeneous data, providing a complete data foundation for subsequent analysis. It effectively eliminates inter-frame displacement errors in video caused by camera shake or rapid target movement. It accurately acquires key static features in the scene, providing a reliable benchmark for spatial modeling. It establishes a quantifiable and analyzable road network spatial relationship model. It uses real motion data to correct the theoretical model, improving the accuracy of spatial representation. It outputs comprehensive monitoring data that integrates spatial topology and real-time dynamic information.

[0054] Furthermore, this application also extracts trajectory position sequences from motion trajectory data and maps them to nodes in a topology graph. For any two adjacent nodes in the graph that are directly connected by a topology edge, when the deviation between the movement direction angle value and the preset direction angle value of the corresponding topology edge exceeds a threshold, the constraint weight value of the topology edge is dynamically updated. Finally, a new topology graph is generated based on the updated weight information.

[0055] Furthermore, this application realizes dynamic graph optimization based on real motion trajectories. Through directional consistency verification and weight adaptive adjustment mechanisms, it effectively corrects the deviation between the theoretical topology model and the actual traffic flow, improving the accuracy of the road network spatial representation and the ability to identify abnormal driving behavior.

[0056] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1A flowchart illustrating a real-time video analysis method based on intelligent image processing, provided in an embodiment of this application;

[0059] Figure 2 A schematic diagram of the structure of a real-time video analysis system based on intelligent image processing provided in this application embodiment;

[0060] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0061] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0062] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0063] Existing multi-view video stitching techniques based on feature point matching have significant shortcomings: in dynamic scenes, the displacement of moving targets can easily lead to stitching misalignment; feature point matching is significantly affected by changes in lighting and occlusion; and relying on offline calibration makes it difficult to adapt to real-time changes in camera pose. Furthermore, incomplete separation of moving targets from the background can cause ghosting, affecting the accuracy of subsequent analysis. These deficiencies stem from insufficient compensation for dynamic offsets between video frames in existing solutions and the lack of a topology optimization mechanism that integrates motion trajectory data, leading to reduced reliability of video analysis results in complex scenes.

[0064] To address the aforementioned issues, this application proposes a real-time video analysis method based on intelligent image processing. This method acquires multi-view video data from a traffic monitoring area and vehicle motion trajectory data. An optical flow compensation algorithm is used to correct inter-frame offset in the multi-view video data, generating corrected multi-view video data. Geometric feature points are extracted from the corrected video data, and a topology graph is constructed. The topology graph is then optimized based on the motion trajectory data, generating an optimized topology graph. Finally, the optimized topology graph and the corrected multi-view video data are combined to generate the video analysis results. This method effectively eliminates inter-frame offset in dynamic scenes through optical flow compensation and optimizes the topology graph using motion trajectory data. It solves problems such as splicing misalignment, illumination interference, and reliance on offline calibration in existing technologies, improving the accuracy and real-time performance of video analysis in complex scenes.

[0065] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0066] Figure 1 A flowchart of a real-time video analysis method based on intelligent image processing provided in this application embodiment is shown below. Figure 1 As shown, the method includes:

[0067] Step 101: Obtain multi-view video data and vehicle motion trajectory data of the traffic monitoring area.

[0068] In step 101, multi-view video data refers to video sequences synchronously collected from multiple cameras in the traffic monitoring area, containing road scene information from different angles. The video frames of each camera are strictly synchronized in time. Motion trajectory data represents continuous vehicle position information obtained through vehicle-mounted positioning devices or roadside sensors, including attributes such as timestamps, latitude and longitude coordinates, and direction of movement.

[0069] In this embodiment, the system first deploys multiple wide-angle cameras to cover the target monitoring area, ensuring that the fields of view of each camera have sufficient overlap; at the same time, it accesses real-time trajectory data transmitted by the vehicle positioning device, establishes a unified timestamp to achieve spatiotemporal alignment between video data and trajectory data; the video acquisition module synchronously captures the images of each camera at a fixed frame rate, the trajectory receiving module parses the vehicle position information in real time, and finally outputs time-aligned multi-source data for subsequent processing.

[0070] For example, at the intersection of a main road in a city, four high-definition cameras are installed to cover the four directions of east, west, north, and south. The cameras simultaneously capture video at 25 frames per second. At the same time, the system receives vehicle GPS data from passing vehicles through roadside units. The data includes vehicle ID, latitude and longitude coordinates, and speed information. The system uses network time protocol to control the time deviation between video frames and trajectory data to the millisecond level, ensuring the spatiotemporal consistency of subsequent processing.

[0071] Step 102: Using an optical flow compensation algorithm, perform inter-frame offset correction on the multi-view video data to generate corrected multi-view video data.

[0072] In step 102, the optical flow compensation algorithm represents a technique for establishing an inter-frame displacement model by analyzing the motion vectors of pixels in adjacent video frames. Inter-frame offset correction refers to eliminating positional deviations between consecutive video frames caused by camera shake or rapid target movement. The corrected multi-view video data refers to the video sequence that has undergone continuous inter-frame offset correction.

[0073] In this embodiment, for each camera's video stream, feature points are extracted from two consecutive frames; the displacement vectors of the feature points in the preceding and following frames are calculated to form an optical flow field; the main motion direction of the optical flow vectors is statistically analyzed, and abnormal offsets are filtered out; an affine transformation model is established based on the effective optical flow vectors, and subsequent frame images are geometrically transformed according to the model parameters to output a video sequence with corrected position.

[0074] For example, in video captured by an east-facing camera, the system detected slight camera shaking due to strong winds between frames 100 and 101. Using an optical flow algorithm, the system calculated the displacement of 10 stable feature points. After excluding two outliers, it averaged the offset of the eight valid points by 5 pixels horizontally and 3 pixels vertically. This was then fused with historical data from the previous five frames (forgetting factor α = 0.6) to obtain smoothing parameters Δx' = 4.8 pixels and Δy' = 2.9 pixels. A translation transformation model was then established to perform reverse compensation on the subsequent 102 frames and beyond, eliminating the effects of shaking. The positions of stationary objects in the compensated video remained stable.

[0075] Step 103: Extract geometric feature points from the corrected multi-view video data, and construct a topological relationship graph based on the geometric feature points.

[0076] In step 103, geometric feature points represent points in the video image with significant geometric characteristics, such as building edge feature points, lane line intersections, and traffic sign corner points. The topological graph is a graph structure with feature points as nodes and the spatial connections between feature points as edges.

[0077] In this embodiment, stable geometric feature points are detected from the corrected video frames; feature points from multiple perspectives are matched and fused to eliminate duplicate points; based on the actual spatial distribution of feature points, points that are close to each other are clustered into key nodes; based on the actual road connection situation, directed edges are established between adjacent nodes, and the direction of the edges represents the legal driving direction of the vehicle, and finally a topological representation of the road network is generated.

[0078] For example, the system extracts 32 lane line intersections and 16 traffic sign corner points from videos from four perspectives; after coordinate transformation and deduplication, it retains 20 spatially unique feature points; based on the actual road layout, it connects these points to form a topology graph containing 15 directed edges, where the edges of east-west main roads are set to bidirectional and side roads are set to one-way. The system reserves a topology graph expansion mechanism to cope with temporary road changes (such as construction).

[0079] Step 104: Optimize the topology graph based on the motion trajectory data to generate an optimized topology graph.

[0080] In step 104, the optimized topology diagram refers to the road network model after correction by motion trajectory data. Nodes represent key traffic points, edges represent travel paths, and the constraint weights of the edges have been adjusted according to the actual traffic flow direction deviation, which can more accurately reflect the real road usage.

[0081] In this embodiment, motion trajectory data is mapped onto nodes of the topology graph; the direction of motion through each edge is counted; the average deviation between the trajectory direction and the preset direction of the edge is calculated; when the deviation exceeds a threshold, the constraint weight of the edge is increased proportionally; after traversing all edges to complete the weight adjustment, unreliable edges with too low weights are deleted, and the optimized topology graph is output.

[0082] For example, the system analyzed 1,000 trajectory data and found that 12% of the preset one-way edges of a certain branch road had reverse driving trajectories with an average directional deviation of 25 degrees. The constraint weight of this edge was increased from 1.0 to 1.8. At the same time, a temporary passage edge with only 3 trajectories was deleted. The final optimized topology map is more in line with the actual traffic conditions.

[0083] Step 105: Generate video analysis results based on the optimized topology graph and the corrected multi-view video data.

[0084] In step 105, the video analysis results represent dynamic analysis data that integrates topology and real-time video, including information such as traffic flow statistics and abnormal event detection.

[0085] In this embodiment, moving targets in real-time video are associated with nodes in the topology graph; the legality of driving is determined based on the target's movement path in the topology graph; traffic flow change trends are predicted by combining historical trajectory data; areas with abnormal changes in topology edge weights are detected as potential event points; and a comprehensive analysis report containing warnings of violations and congestion predictions is output.

[0086] For example, if the system detects that a truck has violated the directional constraints of two one-way streets in a row, it will immediately trigger a wrong-way warning; at the same time, based on the trend of the continuous decrease in the weight of the westbound main road, it will predict that there will be congestion in that direction and generate a warning message to be pushed to the traffic management center.

[0087] This method achieves accurate video analysis in complex traffic scenarios through multi-source data fusion and dynamic optimization mechanisms. Optical flow compensation ensures the stability of video data, topology graphs provide structured road representations, and motion trajectory optimization enables the model to continuously adapt to changes in actual traffic flow. Ultimately, the system can reliably identify traffic violations, accurately predict traffic flow trends, and improve the real-time performance and accuracy of intelligent traffic management.

[0088] To address the dynamic optimization problem of topology graphs in complex traffic scenarios, in some embodiments, step 104: optimizing the topology graph based on the motion trajectory data to generate an optimized topology graph includes:

[0089] Step 201: Extract the trajectory position sequence from the motion trajectory data.

[0090] In step 201, the trajectory position sequence refers to the set of position data of the vehicle at consecutive time points, including timestamps, coordinate information and direction of movement, which is used to reflect the actual driving path of the vehicle.

[0091] In this embodiment, the system extracts continuous and valid location points from the original trajectory data, filters out stationary points and abnormal jump points, and generates a smooth trajectory sequence in chronological order to provide standardized input data for subsequent mapping.

[0092] Step 202: Map each location point in the trajectory location sequence to the corresponding node in the topology graph.

[0093] In step 202, a location point refers to a single coordinate point in the vehicle's trajectory, derived from continuous location data collected by the onboard GPS device, including information such as timestamps and latitude / longitude. A corresponding node refers to the road critical point in the topology graph that is spatially closest to the trajectory location point; the nearest neighbor is determined by calculating the Euclidean distance between the location point and all nodes. The node mapping process refers to matching the trajectory location point to the nearest node in the topology graph. Nodes in the topology graph represent road critical points such as intersection centers or lane line endpoints.

[0094] In this embodiment, the spatial distance between each trajectory location point and all nodes is calculated, and the nearest node is found as the mapping target. When multiple consecutive location points are mapped to the same node, deduplication is performed to ensure that each trajectory forms a coherent path representation in the graph.

[0095] Step 203: For any two adjacent nodes in the topology graph that are directly connected by a topological edge, if the angle deviation between the movement direction angle of the two adjacent nodes and the preset direction angle of the topological edge connecting the two adjacent nodes exceeds a preset angle threshold, then update the constraint weight value of the topological edge.

[0096] In step 203, adjacent nodes refer to two nodes directly connected by an edge in the topology graph, and their adjacency is determined based on the road connection relationships set during the initial graph construction. The movement direction angle value refers to the actual driving direction angle of the vehicle when passing two adjacent nodes, calculated clockwise with true north as the reference. The angle deviation refers to the angle difference between the vehicle's actual movement direction and the preset direction of the topology edge. The preset angle threshold represents the critical value for determining whether the trajectory direction violates regulations, and is preset according to road grade and traffic management requirements. The constraint weight value represents the numerical value of the topology edge direction constraint strength, with an initial value of 1.0, dynamically adjusted based on the trajectory direction deviation.

[0097] In this embodiment, all trajectory directions of each topological edge are statistically analyzed, and the average deviation between the deviation and the preset direction of the edge is calculated. When the deviation continues to exceed the threshold, the constraint weight value of the edge is increased proportionally to enhance the ability to identify abnormal driving behavior.

[0098] Step 204: Based on the updated constraint weight values, optimize the weight information of the topological edges in the topological graph to generate an optimized topological graph.

[0099] In step 204, the optimization process refers to the process of adjusting the connection relationships of the topology graph according to the updated constraint weights. Weight information refers to the set of constraint weight values ​​for all edges in the topology, used to characterize the strictness of the enforcement of the access rules for each path.

[0100] In this embodiment, the system traverses all topological edges, removes unreliable connections with weights below the critical value, merges redundant edges with similar directions, and finally generates a simplified and reliable optimized topological graph.

[0101] Here is a specific example:

[0102] In the optimization process of a main road intersection in a city, the system first extracts continuous trajectory position sequences from the received vehicle GPS data, filters out stationary points with dwell times exceeding a threshold and points with abrupt coordinate changes, and obtains the effective driving paths of vehicles. The cleaned trajectory position points are mapped to nodes in a pre-constructed topological graph. The preset direction angles for the east-west main road's bidirectional edges are 90 degrees and 270 degrees, respectively, while the preset direction for the northbound branch road's unidirectional edge is 180 degrees. Statistical analysis of 1000 effective trajectory data points reveals that 120 trajectories on the northbound branch road exhibit reverse driving behavior. The calculated average direction angle of these trajectories is 0 degrees, deviating 180 degrees from the preset direction of 180 degrees. Based on the preset angle threshold of 30 degrees, this deviation significantly exceeds the threshold. The system adjusts the constraint weight value of this edge according to the formula W_new = W_old × 1.5, where W_old is the original weight value of 1.0, and W_new is updated to 1.5. Meanwhile, it was detected that a temporary edge connecting a branch road and a parking lot was used by only 3 trajectories, which was less than the set proportion threshold of the total number of trajectories. The system removed the edge from the topology graph. After optimization, the new topology graph accurately reflects the actual rules of two-way traffic on the main road and one-way traffic on the branch road, and strengthens the constraints on high-frequency violation sections.

[0103] In this embodiment, a dynamic optimization mechanism driven by trajectory data enables the topology graph to continuously adapt to changes in actual traffic flow, effectively improving the reliability of path compliance analysis in complex scenarios while maintaining a rapid response capability to changes in the road network.

[0104] To address the dynamic optimization problem of directional constraints in topological graphs, in some embodiments, step 203—when the angular deviation between the movement direction angle value of the two adjacent nodes and the preset direction angle value of the topological edge connecting the two adjacent nodes exceeds a preset angle threshold—involves updating the constraint weight value of the topological edge, including:

[0105] Step 301: Calculate the movement direction angle value of the two adjacent nodes, and use the movement direction angle value as the first direction value.

[0106] In step 301, the calculation of the movement direction angle value means calculating the direction vector using the coordinates of the trajectory points between adjacent nodes and taking the arctangent to obtain the angle. For example, if a vehicle passes through nodes A (100,50) and B (105,55), with a direction vector of (5,5), the calculated arctan(5 / 5) = 45 degrees. The first direction value refers to the actual movement direction angle of the vehicle when passing through two adjacent nodes, which is calculated using the trajectory coordinates.

[0107] In this embodiment of the application, the system extracts continuous position points that pass through the two adjacent nodes from the trajectory position sequence, calculates the vehicle's movement direction based on the changes in the position point coordinates, and converts it into a standard angle representation.

[0108] Step 302: Use the angle value of the preset direction of the topological edge connecting the two adjacent nodes as the second direction value.

[0109] In step 302, the preset direction angle value refers to the legal driving direction angle of the road pre-set in the topology graph, reflecting the vehicle travel direction required by traffic rules. The second direction value refers to the preset legal travel direction angle of the topology edge, which is set according to traffic rules during graph construction.

[0110] In this embodiment of the application, the system reads the direction parameters stored in the topological edge attributes, which are set according to the actual traffic rules of the road during the initial mapping.

[0111] Step 303: Calculate the absolute difference between the first direction value and the second direction value.

[0112] In step 303, the absolute difference refers to the minimum angular difference between the first direction value and the second direction value, which is used to quantify the degree of deviation between the actual driving direction and the specified direction.

[0113] In this embodiment of the application, the system calculates the difference between two angle values ​​and takes the absolute value. If the result is greater than 180 degrees, the value is subtracted from 360 degrees to ensure that the minimum angle difference is obtained.

[0114] Step 304: If the absolute difference exceeds the preset angle threshold, the constraint weight value of the topological edge connecting the two adjacent nodes is multiplied by the incremental coefficient to generate an updated constraint weight value.

[0115] In step 304, the incremental coefficient is the amplification factor used to dynamically adjust the constraint weight value, and different coefficients are set according to the road importance level. The updated constraint weight value represents the adjusted edge direction constraint strength; the larger the value, the stricter the monitoring of that edge direction.

[0116] In this embodiment of the application, the system maintains a coefficient lookup table. When an angle deviation exceeding the limit is detected, the corresponding coefficient is selected according to the road grade for weight adjustment.

[0117] Here is a specific example:

[0118] In a traffic monitoring system at a major intersection in a city, the system dynamically optimizes the one-way traffic rules for northbound side roads. First, it acquires 120 reverse-traffic trajectories passing through the road segment, calculates the angle of movement direction when each trajectory passes adjacent nodes, and takes the average to obtain a first direction value of 0 degrees. This value is derived from the arctangent calculation of the coordinates of the trajectory's start and end points. Simultaneously, it reads the preset direction angle value of 180 degrees from the side road's topology as the second direction value. The system calculates the absolute difference between the two, obtaining 180 degrees, and uses the formula Δθ=min|θ1-θ2|,360-|θ1-θ2| for normalization, ultimately determining the actual angle deviation to be 180 degrees. Since this value exceeds the preset angle threshold of 30 degrees, the system selects an increment coefficient of 1.5 based on the side road level and updates the constraint weight value according to the formula W_new=W_old×k, where W_old is the original weight of 1.0 and k is the increment coefficient of 1.5, resulting in a new weight value of 1.5.

[0119] In this embodiment, a dynamic direction comparison and weight adjustment mechanism enables the topology graph to adapt to changes in traffic flow. While maintaining the rigidity of road rules, it enables key monitoring of high-frequency violation sections, effectively improving the level of intelligence in traffic management.

[0120] To improve the dynamic adaptability and analysis accuracy of traffic monitoring systems, in some embodiments, step 105: generating video analysis results based on the optimized topology diagram and the corrected multi-view video data includes:

[0121] Step 401: Match the geometric feature points in the corrected multi-view video data with the nodes in the optimized topology graph.

[0122] In step 401, the location matching process refers to the process of comparing the spatial locations of the geometric feature points detected in the video data with the spatial locations of existing nodes in the topological relationship graph.

[0123] In this embodiment, the system calculates the spatial distance between the coordinates of video feature points and the coordinates of topological nodes. When the distance is less than the matching threshold, it is determined that the match is successful; otherwise, it enters the process of adding a new node.

[0124] Step 402: If the matching fails, the geometric feature point corresponding to the failed matching is determined as a new geometric feature point, and the minimum Euclidean distance between the new geometric feature point and all existing nodes is calculated.

[0125] In step 402, newly added geometric feature points refer to road feature points that newly appear in the video and are not registered in the topology map. Existing nodes refer to road key points that already exist in the topology map, derived from historical feature point clustering results. The minimum Euclidean distance is the shortest straight-line distance between this point and all nodes in the topology map.

[0126] In this embodiment of the application, the system calculates the square root of the sum of the squares of the coordinate differences between the new feature point and each existing node, and records the minimum value as the basis for judgment.

[0127] Step 403: When the minimum Euclidean distance exceeds the preset distance threshold, insert a new node corresponding to the newly added geometric feature point in the topology graph and update the connection relationship set.

[0128] In step 403, the preset distance threshold refers to the critical distance value (e.g., 5 meters) used to determine whether to add a new node, and is set according to road density and monitoring accuracy requirements. A new node represents a newly discovered road feature point in the video, which is added to the topology map after distance verification. The process of inserting a new node involves assigning a unique ID to the new feature point, setting coordinate attributes, and searching for neighboring nodes to establish connections.

[0129] In this embodiment, when the minimum Euclidean distance does not exceed the threshold, the newly added feature point is classified into the feature set of the nearest node, and no new node is created. The system uses the new node as the center, searches for existing nodes within the radius, establishes bidirectional or unidirectional connection edges according to the actual road connection situation, and initializes the weight values.

[0130] Step 404: Based on the motion trajectory data, verify whether the connection path between the new node and all existing nodes is valid.

[0131] In step 404, a connected path refers to a valid sequence of connections between nodes in the topology graph. The connected path verification process is the process of verifying whether the motion trajectory supports the newly established topological connections.

[0132] In this embodiment of the application, the system retrieves historical trajectory data, finds trajectory records that pass through both new nodes and adjacent nodes, and calculates the degree of matching between their movement direction and the topological edge direction.

[0133] Step 405: If the verification finds that at least one path of the motion trajectory is inconsistent with the preset connected path in the updated topology graph, it is determined that there is a conflict. Then, the constraint weight values ​​of the relevant topology edges are adjusted to obtain the adjusted topology graph.

[0134] In step 405, inconsistent connectivity paths refer to situations where the edge actually traversed by the trajectory does not exist in the graph, or the trajectory direction deviates from the preset edge direction by more than a threshold. "Inconsistency" refers to two situations: first, the edge actually traversed by the trajectory does not exist in the topology graph, such as a vehicle traveling directly from node A to node C, but the graph only shows the ABC path; second, the angle deviation between the trajectory direction and the preset edge direction exceeds a threshold, such as an edge preset to run east-only, while the trajectory runs west-only. Connectivity path conflict determination refers to a non-negligible deviation between the actual trajectory movement direction and the preset topology edge direction. Relevant topology edges refer to edges actually traversed by the trajectory but not present in the graph, as well as edges between new nodes and neighboring nodes. The adjusted topology graph refers to the road network model after verification by newly added nodes and correction of conflicting edges.

[0135] In this embodiment, the system calculates the proportion and deviation angle of the conflict trajectory, adjusts the constraint weight of the relevant edge according to preset rules, and may delete the connection edge when there is a severe conflict.

[0136] Step 406: Generate video analysis results based on the corrected multi-view video data, the adjusted topology diagram, and the motion trajectory data.

[0137] In this embodiment, the system integrates information such as violation identification, congestion prediction, and road network change detection into a structured report.

[0138] Here is a specific example:

[0139] During road construction at a major intersection in a city, the system, while processing corrected multi-view video data in real time, discovered a new corner feature point of a temporary construction fence in the field of view of the east-facing camera. The coordinates of this feature point were matched with 20 existing nodes in the optimized topology graph. The Euclidean distance between this feature point and the nearest node, the east-facing pedestrian crossing sign, was calculated as d = sqrt[(x2-x1)² + (y2-y1)²] = 6.3 meters, where x1 and y1 are the coordinates of the pedestrian crossing node (e.g., (x1, y1) = (0, 0)), and x2 and y2 are the coordinates of the new feature point (e.g., (x2, y2) = (6.3, 0)). Since this distance exceeded a preset 5-meter threshold, the system determined it to be a newly added geometric feature point and created a corresponding new node in the topology graph. A search was then conducted within a 7-meter radius of this node, encompassing the east-facing pedestrian crossing node and the southeast corner streetlight node. Based on actual road conditions, the system establishes bidirectional connections between the new node and the pedestrian crossing node, with an initial weight of 1.0; simultaneously, it establishes unidirectional connections with the street light node. The system then backtracks and analyzes the most recent 1000 motion trajectory data, finding that 35 trajectories support connectivity between the new node and the pedestrian crossing node, and the directions are consistent; however, only 2 trajectories support connections with the street light node, and their directions are inconsistent. Based on the verification results, the system deletes the connection edges with the street light node and increases the weight of the connection edges with the pedestrian crossing node to 1.2, resulting in an adjusted topology graph. Based on this adjusted topology graph, the system monitors in real time a construction vehicle illegally entering the pedestrian crossing area from the new node (directional deviation reaching 65 degrees, exceeding the 30-degree threshold), immediately generating a construction vehicle violation alarm; simultaneously, combined with trajectory data stream analysis, it finds that the traffic speed on the eastbound main road has decreased by 20%, predicting congestion within 10 minutes, and automatically sends a warning message containing specific location coordinates and recommended detour routes to the traffic management center. The final video analysis output includes: First, real-time violation reports (vehicle ID, violation type, timestamp); second, dynamic traffic prediction (congested areas, estimated duration); and third, topology map change records (coordinates of newly added nodes, details of edge adjustments).

[0140] In this embodiment, by dynamically adding or deleting nodes and verifying connection relationships, the system can adapt to changes in the road network, ensuring that the traffic analysis results are always consistent with the actual situation, while improving the accuracy of identifying violations and abnormal conditions.

[0141] To improve the dynamic update mechanism of the topology graph in the traffic monitoring system, in some embodiments, step 403: inserting the new node corresponding to the newly added geometric feature point into the topology graph and updating the connection set includes:

[0142] Step 501: Obtain the set of neighboring nodes within the preset radius of the new node.

[0143] In step 501, the neighboring node set refers to the set of all road key node nodes that already exist in the topology graph within a circular area centered on the new node and with a preset radius.

[0144] In this embodiment of the application, the system quickly retrieves existing nodes that meet the distance conditions through spatial indexing. These nodes represent traffic element feature points such as intersections and lane line endpoints.

[0145] Step 502: Create a new node in the topology graph and set the position coordinates of the new node to the coordinates of the newly added geometric feature point.

[0146] In step 502, the position coordinates of the new node refer to the spatial position data of the newly added geometric feature point in the video coordinate system.

[0147] In this embodiment, the system maps the coordinates of feature points detected in the video to a mapping with topological relationships using pre-calibrated transformation parameters. Figure 1 In a unified global coordinate system, ensure consistency of spatial position.

[0148] Step 503: Establish a connection between the new node and each node in the set of neighboring nodes, and add the connection to the set of connection relationships.

[0149] In step 503, the connection establishment process refers to determining the connection direction and type (one-way / two-way) between the new node and its neighboring nodes according to the actual road traffic rules, and initializing the weight values. The connection addition process refers to adding the newly created edge object (including direction and weight attributes) to the edge set data structure of the topology graph.

[0150] In this embodiment of the application, the system establishes directional and weighted connection edges between the new node and each neighboring node based on the road type and traffic control information, and initializes the constraint weight values ​​of these edges.

[0151] Here is a specific example:

[0152] During road construction at a major intersection in a city, the system detected a new corner feature point of a temporary construction fence using an eastward-facing camera. Its coordinates, after conversion, were determined to be x=125.4 meters and y=78.6 meters. Using this point as the center, the system searched for neighboring nodes within a preset 7-meter radius. The coordinates of the eastward-facing pedestrian crossing node were calculated to be x=126.0 meters and y=78.2 meters, with a distance d=sqrt[(126.0-125.4)²+(78.2-78.6)²]=0.72 meters; the coordinates of the southeast corner streetlight node were x=126.2 meters and y=78.8 meters, with a distance d=sqrt[(126.2-125.4)²+(78.8-78.6)²]=0.82 meters, both within the search range. After creating new nodes and setting their coordinates, the system determines, based on the construction plan design drawings, that temporary fencing and pedestrian crossings should maintain two-way traffic. Therefore, a two-way connection edge is established with an initial weight of 1.0. Due to physical isolation from street light nodes, only a one-way connection edge is established for monitoring purposes. New connection relationships are recorded in the connection relationship set. The preset directions for the two-way connection edges are set to 120 degrees from the fencing to the pedestrian crossing and 300 degrees in the opposite direction, respectively. The one-way connection edge is set to 60 degrees from the fencing to the street light.

[0153] In this embodiment, a standardized process for inserting new nodes and establishing connections ensures that the topology graph can respond promptly to changes in road layout, maintains the traffic monitoring system's ability to accurately model dynamic environments, and provides reliable support for real-time traffic analysis and decision-making.

[0154] To establish a topological model that accurately reflects the road network structure, in some embodiments, step 103: constructing a topological relationship graph based on the geometric feature points includes:

[0155] Step 601: Based on the spatial location of the geometric feature points, cluster the geometric feature points corresponding to the same physical location into a location feature set.

[0156] In step 601, spatial location refers to the coordinate values ​​of geometric feature points in a unified coordinate system, obtained through camera calibration parameter transformation. The same physical location refers to multiple video feature points corresponding to the same traffic element (such as the center of an intersection) in the real world. The location feature set refers to the clustering result of geometric feature points belonging to the same traffic element in spatial location.

[0157] In this embodiment of the application, the system uses a spatial density clustering algorithm to group geometric feature points such as lane line intersections and traffic sign corners that are close in distance and have similar features into a feature set representing key points of the same intersection or road.

[0158] Step 602: Generate a set of connection relationships based on the spatial adjacency relationships between adjacent sets of location features.

[0159] In step 602, the set of connection relationships refers to the road connectivity relationships between different location feature sets, as well as the sum of the connection relationships between all nodes in the topology graph, including attributes such as edge direction and weight.

[0160] In this embodiment of the application, the system analyzes the spatial distribution and road direction among the feature sets, establishes connection relationships between feature sets with actual passage possibilities, and records the connection direction and initial weight.

[0161] Step 603: Construct a topological graph using the set of location features as nodes and the set of connection relationships as topological edges.

[0162] In this embodiment, the system uses the clustered feature set as graph nodes, establishes weighted directed edges based on the connection relationship set, and finally forms a topological model that fully describes the road network.

[0163] Here is a specific example:

[0164] In the implementation of an intelligent transportation system at a major intersection in a city, the system processes stable video data collected by four cameras. First, eight lane intersections are extracted from the east-facing cameras, with coordinates from (x1, y1) to (x8, y8). Seven intersections are extracted from the west, six from the south, and five from the north. After coordinate system transformation, a density clustering algorithm is used to group feature points with a spatial distance of less than 2 meters into location feature sets, ultimately forming four main sets: the east-facing approach set contains three feature points, with the center coordinates calculated using the arithmetic mean (125.6, 78.3); the west-facing approach set contains two feature points, with the center at (75.4, 78.5); the south-facing approach set contains three feature points, with the center at (100.2, 55.1); and the north-facing approach set contains two feature points, with the center at (100.5, 101.7). Based on the actual road design drawings, the system establishes bidirectional connecting edges between the east-west junctions, with directional angles of 90 degrees and 270 degrees respectively; bidirectional connecting edges between the south-north junctions, with directional angles of 0 degrees and 180 degrees; and unidirectional connecting edges between each approach lane and its corresponding turning lane junction, such as the east approach lane to south turning lane junction with a directional angle of 210 degrees. The initial weight of each edge is set to 1.0, with an additional 0.2 weight added to turning edges. The final constructed topology graph contains 4 nodes and 8 edges, accurately reflecting the actual traffic rules of the intersection.

[0165] In this embodiment, intelligent clustering of geometric feature points and construction of topological relationships enable the automatic conversion from video data to a road network model, providing a structured road network foundation for subsequent traffic analysis and anomaly detection.

[0166] To improve the stability of video data, in some embodiments, step 102: using an optical flow compensation algorithm to correct inter-frame offset in the multi-view video data and generate corrected multi-view video data includes:

[0167] Step 701: Calculate the inter-frame motion vector between adjacent video frames in the multi-view video data using an optical flow compensation algorithm.

[0168] In step 701, adjacent video frames refer to two frames with consecutive timestamps, and their adjacency is determined by the frame number order. The inter-frame motion vector refers to the displacement direction and magnitude of corresponding feature points in adjacent video frames.

[0169] In this embodiment of the application, the system analyzes the positional changes of feature points of a fixed object in two consecutive frames of images, calculates the amount of movement of each feature point in the horizontal and vertical directions, and forms a vector field describing the overall inter-frame motion.

[0170] Step 702: Filter the set of valid motion regions from the inter-frame motion vectors.

[0171] In step 702, the screening process for effective motion regions involves statistically analyzing the direction histograms of all motion vectors, retaining regions whose directions fall within ±15 degrees of the main peak, and excluding isolated abnormal regions. The set of effective motion regions refers to continuous image regions with consistent motion directions, and their effectiveness is determined by directional consistency and spatial continuity.

[0172] In this embodiment, the system statistically analyzes the directional distribution of all inter-frame motion vectors, marks regions whose directional deviation from the mainstream direction is less than a threshold as valid, and excludes abnormal motion regions caused by local occlusion or noise.

[0173] Step 703: Recursively accumulate the offset of the effective moving area set, and calculate the inter-frame offset correction parameter based on the accumulation result.

[0174] In step 703, the offset refers to the positional difference between video frames, calculated using the effective movement region vector. The recursive accumulation process refers to the weighted fusion of the effective offset of the current frame with historical offsets. The inter-frame offset correction parameter refers to the amount of displacement that the video frame needs to compensate for.

[0175] In this embodiment, the system uses a sliding window mechanism to perform a weighted average of the effective offsets of the most recent frames, smoothing out instantaneous fluctuations and obtaining stable inter-frame offset correction parameters.

[0176] Step 704: Adjust the position coordinates of the video frame according to the inter-frame offset correction parameters.

[0177] In step 704, the position coordinates refer to the position of the video pixel in the image coordinate system, which is obtained through feature point detection. The process of adjusting the position coordinates refers to the process of performing geometric transformations on the video frame according to the correction parameters.

[0178] In this embodiment, the system establishes an affine transformation model, converts the correction parameters into a transformation matrix, and performs reverse displacement compensation on the video frames.

[0179] Step 705: Based on the adjusted video frames, correct the multi-view video data to generate corrected multi-view video data.

[0180] In step 705, the video data correction process refers to applying the calculated geometric transformation parameters (such as translation matrix) to each frame of the image and resampling to generate a stable video sequence.

[0181] In this embodiment, the system applies correction parameters to all video frames sequentially to ensure that the static background remains stable throughout the entire video sequence.

[0182] Here is a specific example:

[0183] In a surveillance scenario at a main intersection in a city, an east-facing camera exhibited periodic shaking due to strong winds. When processing video data from frames 100 to 101, the system selected 10 stable feature points, including traffic sign corners and streetlight bases. The system calculated the movement vectors of each point using an optical flow algorithm. Eight of these feature points were found to be moving downwards and to the right, with an average horizontal offset of Δx = 5 pixels and a vertical offset of Δy = 3 pixels. The calculation formulas are: Δx = 1 / n∑(xi_101 - xi_100), Δy = 1 / n∑(yi_101 - yi_100), where n = 8 is the number of valid feature points, and xi and yi are the coordinates of the feature points. After excluding two abnormally moving feature points, the system weighted and fused the offsets of the eight valid moving areas with historical data from the previous five frames. A forgetting factor of α = 0.6 was used for recursive calculation, resulting in smoothed correction parameters Δx' = 4.8 pixels and Δy' = 2.9 pixels. A translation transformation matrix M=[10-Δx';01-Δy'] is established to perform geometric correction on the 102nd frame and subsequent video frames.

[0184] In this embodiment, the optical flow compensation and dynamic correction mechanism effectively eliminates video instability caused by external interference such as camera shake, ensuring the accuracy of subsequent topology construction and traffic analysis, and improving the system's reliability in complex environments.

[0185] Figure 2 A schematic diagram of the structure of a real-time video analysis system based on intelligent image processing provided in this application embodiment is shown below. Figure 2 As shown, the system includes:

[0186] The acquisition module 21 is used to acquire multi-view video data and vehicle motion trajectory data of the traffic monitoring area.

[0187] The first generation module 22 is used to perform inter-frame offset correction on the multi-view video data using an optical flow compensation algorithm to generate corrected multi-view video data.

[0188] The construction module 23 is used to extract geometric feature points from the corrected multi-view video data and construct a topological relationship graph based on the geometric feature points.

[0189] The second generation module 24 is used to optimize the topology graph based on the motion trajectory data and generate an optimized topology graph.

[0190] The third generation module 25 is used to generate video analysis results based on the optimized topology diagram and the corrected multi-view video data.

[0191] Figure 2 The aforementioned real-time video analysis system based on intelligent image processing can perform... Figure 1 The implementation principle and technical effects of the real-time video analysis method based on intelligent image processing described in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the real-time video analysis system based on intelligent image processing in the above embodiments are performed have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0192] In one possible design, Figure 2 The real-time video analysis system based on intelligent image processing shown in the embodiment can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0193] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0194] The processing component 32 is used to perform the above. Figure 1 The embodiment describes a real-time video analysis method based on intelligent image processing.

[0195] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0196] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0197] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0198] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0199] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0200] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0201] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown is a real-time video analysis method based on intelligent image processing.

[0202] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0203] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0204] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0205] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A real-time video analysis method based on intelligent image processing, characterized in that, include: Acquire multi-view video data and vehicle trajectory data of the traffic monitoring area; Using an optical flow compensation algorithm, inter-frame offset correction is performed on the multi-view video data to generate corrected multi-view video data; Geometric feature points are extracted from the corrected multi-view video data, and a topological relationship graph is constructed based on the geometric feature points. The topology graph is optimized based on the motion trajectory data to generate an optimized topology graph; Based on the optimized topology diagram and the corrected multi-view video data, video analysis results are generated. The step of optimizing the topology graph based on the motion trajectory data to generate an optimized topology graph includes: Extract the trajectory position sequence from the motion trajectory data; Map each location point in the trajectory location sequence to the corresponding node in the topological relationship graph; For any two adjacent nodes in the topological graph that are directly connected by a topological edge, if the angle deviation between the movement direction angle of the two adjacent nodes and the preset direction angle of the topological edge connecting the two adjacent nodes exceeds a preset angle threshold, then the constraint weight value of the topological edge is updated. Based on the updated constraint weight values, the weight information of the topological edges in the topological graph is optimized to generate an optimized topological graph. The process of generating video analysis results based on the optimized topology diagram and the corrected multi-view video data includes: The geometric feature points in the corrected multi-view video data are matched with the nodes in the optimized topology graph. If a match fails, the geometric feature point corresponding to the failed match is determined as a new geometric feature point, and the minimum Euclidean distance between the new geometric feature point and all existing nodes is calculated. When the minimum Euclidean distance exceeds a preset distance threshold, a new node corresponding to the newly added geometric feature point is inserted into the topology graph, and the connection set is updated. Based on the motion trajectory data, verify whether the connection path between the new node and all existing nodes is valid; If the verification finds that at least one path of the motion trajectory is inconsistent with the preset connected path in the updated topology graph, it is determined that there is a conflict. Then the constraint weight values ​​of the relevant topology edges are adjusted to obtain the adjusted topology graph. Based on the corrected multi-view video data, the adjusted topology diagram, and the motion trajectory data, video analysis results are generated.

2. The method according to claim 1, characterized in that, When the angle deviation between the movement direction angle of the two adjacent nodes and the preset direction angle of the topological edge connecting the two adjacent nodes exceeds a preset angle threshold, the constraint weight value of the topological edge is updated, including: Calculate the movement direction angle value of the two adjacent nodes, and use the movement direction angle value as the first direction value; The angle value of the preset direction of the topological edge connecting the two adjacent nodes is used as the second direction value; Calculate the absolute difference between the first direction value and the second direction value; If the absolute difference exceeds the preset angle threshold, the constraint weight value of the topological edge connecting the two adjacent nodes is multiplied by the incremental coefficient to generate an updated constraint weight value.

3. The method according to claim 1, characterized in that, The step of inserting the new node corresponding to the newly added geometric feature point into the topological graph and updating the connection set includes: Obtain the set of neighboring nodes within a preset radius of the new node; Create a new node in the topology graph, and set the position coordinates of the new node to the coordinates of the newly added geometric feature point; Establish connections between the new node and each node in the set of neighboring nodes, and add the connections to the set of connections.

4. The method according to claim 1, characterized in that, The construction of the topological relationship graph based on the geometric feature points includes: Based on the spatial location of the geometric feature points, the geometric feature points corresponding to the same physical location are clustered into a set of location features; Generate a set of connection relationships based on the spatial adjacency relationships between adjacent sets of location features; A topological graph is constructed using the set of location features as nodes and the set of connection relationships as topological edges.

5. The method according to claim 1, characterized in that, The step of using an optical flow compensation algorithm to correct inter-frame offset in the multi-view video data and generate corrected multi-view video data includes: Using an optical flow compensation algorithm, the inter-frame motion vector between adjacent video frames in the multi-view video data is calculated; Filter the set of valid movement regions from the inter-frame motion vectors; The offset of the effective mobile region set is recursively accumulated, and the inter-frame offset correction parameter is calculated based on the accumulation result; Adjust the position coordinates of the video frame according to the inter-frame offset correction parameters; Based on the adjusted video frames, the multi-view video data is corrected to generate corrected multi-view video data.

6. A real-time video analysis system based on intelligent image processing, characterized in that, include: The acquisition module is used to acquire multi-view video data and vehicle motion trajectory data of the traffic monitoring area; The first generation module is used to perform inter-frame offset correction on the multi-view video data using an optical flow compensation algorithm, and generate corrected multi-view video data. The construction module is used to extract geometric feature points from the corrected multi-view video data and construct a topological relationship graph based on the geometric feature points; The second generation module is used to optimize the topology graph based on the motion trajectory data and generate an optimized topology graph. The third generation module is used to generate video analysis results based on the optimized topology diagram and the corrected multi-view video data; The step of optimizing the topology graph based on the motion trajectory data to generate an optimized topology graph includes: Extract the trajectory position sequence from the motion trajectory data; Map each location point in the trajectory location sequence to the corresponding node in the topological relationship graph; For any two adjacent nodes in the topological graph that are directly connected by a topological edge, if the angle deviation between the movement direction angle of the two adjacent nodes and the preset direction angle of the topological edge connecting the two adjacent nodes exceeds a preset angle threshold, then the constraint weight value of the topological edge is updated. Based on the updated constraint weight values, the weight information of the topological edges in the topological graph is optimized to generate an optimized topological graph. The process of generating video analysis results based on the optimized topology diagram and the corrected multi-view video data includes: The geometric feature points in the corrected multi-view video data are matched with the nodes in the optimized topology graph. If a match fails, the geometric feature point corresponding to the failed match is determined as a new geometric feature point, and the minimum Euclidean distance between the new geometric feature point and all existing nodes is calculated. When the minimum Euclidean distance exceeds a preset distance threshold, a new node corresponding to the newly added geometric feature point is inserted into the topology graph, and the connection set is updated. Based on the motion trajectory data, verify whether the connection path between the new node and all existing nodes is valid; If the verification finds that at least one path of the motion trajectory is inconsistent with the preset connected path in the updated topology graph, it is determined that there is a conflict. Then the constraint weight values ​​of the relevant topology edges are adjusted to obtain the adjusted topology graph. Based on the corrected multi-view video data, the adjusted topology diagram, and the motion trajectory data, video analysis results are generated.

7. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement a real-time video analysis method based on intelligent image processing as described in any one of claims 1 to 5.

8. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements a real-time video analysis method based on intelligent image processing as described in any one of claims 1 to 5.

Images