Charging pile oil vehicle occupation prevention intelligent monitoring management system based on Internet of Things video monitoring

Abstract

The invention belongs to the technical field of internet-of-things video monitoring, and provides an intelligent monitoring management system for preventing oil vehicle occupation of a charging pile based on internet-of-things video monitoring, and the system comprises an intelligent monitoring management module which is used for solving the problems of trajectory tracking interruption and occupation prevention failure caused by similar characteristics of a snowfield and a white vehicle body oil vehicle in snowy weather. The method comprises the following steps: acquiring a video through a camera, judging track breakage, analyzing snowfield and vehicle body feature interference, improving the discrimination degree by means of brightness adjustment, LBP texture filtering, shadow and edge enhancement and the like, reconstructing a vehicle body contour by fusing multiple features, and recovering a breakage track through datum point positioning, motion parameter calculation and inertia completion. And finally, inputting an AI model to predict a driving intention and linking a parking lock. The anti-occupation technical bottleneck in special weather is effectively solved, the accuracy and stability of monitoring control are improved, and efficient utilization of the charging pile is guaranteed.

Description

Intelligent Monitoring and Management System for Preventing Fuel-Occupying Charging Stations Based on IoT Video Surveillance Technical Field

[0001] This invention belongs to the field of Internet of Things (IoT) video surveillance technology, specifically an intelligent monitoring and management system for preventing fuel vehicles from occupying charging pile spaces based on IoT video surveillance. Background Technology

[0002] With the development of the new energy vehicle industry and the continuous improvement of charging infrastructure, the problem of gasoline vehicles illegally occupying charging spaces is prominent, hindering the industry's development. Therefore, charging station anti-gasoline vehicle obstruction systems based on IoT video surveillance have been widely applied.

[0003] The core principle of the existing system is as follows: video data is collected through IoT cameras, vehicle outlines are extracted using computer vision technology, vehicle video trajectories are tracked and vehicle driving intentions are predicted using artificial intelligence technology, and when a vehicle (gasoline vehicle) is determined to occupy a charging space, a linkage mechanism is triggered to raise the ground lock, which can effectively intercept vehicles under normal weather conditions.

[0004] However, the system has obvious defects in heavy snow weather: snow accumulation forms a large area of ​​white background, and white cars and oil cars have similar background features, which makes it impossible for computer vision algorithms to effectively distinguish them, and vehicle contour recognition fails (contour is missing or blends into the background).

[0005] Contour recognition failure directly leads to the interruption of vehicle trajectory tracking, making it impossible to extract stable feature points to form a continuous trajectory, paralyzing the driving direction prediction function, and failing to trigger the ground lock linkage mechanism in time. Ultimately, the fuel vehicle successfully occupies the space, and the anti-occupancy function fails.

[0006] Existing technologies are not optimized for this scenario and lack solutions to enhance the contrast between the vehicle body and the snow and to complete the broken trajectory. They cannot meet the actual needs and there is an urgent need for stable recognition and prediction technologies adapted to this special scenario.

[0007] To address this, the present invention provides an intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles, based on Internet of Things video surveillance. Summary of the Invention

[0008] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0009] The technical solution adopted by this invention to solve its technical problem is as follows: This invention proposes an intelligent monitoring and management system for charging piles to prevent oil vehicle occupancy based on IoT video surveillance, including: a trajectory breakage detection module: acquiring vehicle monitoring video and determining whether there is a vehicle trajectory breakage phenomenon in the vehicle monitoring video; an interference analysis and feature enhancement module: if there is a vehicle trajectory breakage phenomenon in the vehicle monitoring video, analyzing whether there is mutual interference between the snow background features and the vehicle body features in the vehicle monitoring video; if there is mutual interference, processing the vehicle monitoring video to enhance the distinguishability between the snow background features and the vehicle body features; a vehicle body contour reconstruction module: based on the vehicle monitoring video after enhancing the distinguishability between the snow background features and the vehicle body features, fusing the vehicle body contour in the vehicle monitoring video to generate a vehicle body feature contour; and a trajectory restoration and intent prediction module: based on the generated vehicle body feature contour, restoring the trajectory breakage monitoring video segment to generate a complete trajectory point sequence for predicting the vehicle's driving intent.

[0010] As a further improvement of the present invention, the specific process of obtaining vehicle monitoring video is as follows: continuously capturing video streams of the area where the vehicle is located in real time through a camera, segmenting and marking the video streams according to the vehicle verification timestamp and camera ID, and obtaining complete vehicle monitoring video.

[0011] As a further improvement of the present invention, the specific process for determining whether there is a vehicle trajectory breakage phenomenon in the vehicle monitoring video is as follows: extract the motion pixel cluster center and high texture entropy value feature points from the vehicle monitoring video; use the Hungarian algorithm to match the motion pixel cluster center and high texture entropy value feature points of the same type between adjacent frames; count the feature point matching rate of 3 consecutive frames and record the number of consecutive frames without effective feature point association; if the feature point matching rate of 2 or more consecutive frames is lower than a preset threshold and the number of frames without effective feature point association is ≥2, then it is determined that there is a vehicle trajectory breakage phenomenon in the vehicle monitoring video, and the video time interval corresponding to the vehicle trajectory breakage phenomenon is marked.

[0012] As a further improvement of the present invention, the specific process of analyzing whether there is mutual interference between snow background features and vehicle body features in vehicle monitoring video is as follows: extract the vehicle candidate region corresponding to the trajectory break within the video time interval, and select the snow background region corresponding to the vehicle candidate region; calculate the RGB color similarity, brightness variance, and brightness variance of the vehicle candidate region and the snow background region respectively; if the color similarity is ≥90%, the brightness variance is ≤15, and the texture entropy difference is ≤0.5, and all three conditions are met, it is determined that there is mutual interference between snow background features and vehicle body features in the vehicle monitoring video.

[0013] As a further improvement of the present invention, the specific process of enhancing the distinction between snow background features and vehicle body features is as follows: extracting moving pixel regions, filtering out moving pixel regions that have generated displacement, brightening the original moving pixel regions by 1.2 times, and attenuating the brightness of the static snow background to 0.8 times the original brightness; extracting LBP high texture regions, performing local binary mode texture filtering on the vehicle monitoring video frame by frame, amplifying the dense texture details of the vehicle body, and suppressing the uniform fluffy texture of the snow.

[0014] As a further improvement of the present invention, the specific process of enhancing the distinction between snow background features and vehicle body features further includes: extracting weak shadow boundary regions, converting vehicle monitoring video clip frames to HSV color space, adjusting the gain of the brightness channel to 1.3 times the original brightness channel, and amplifying the weak shadow band formed by shading at the bottom of the vehicle; extracting edge gradient regions, using the Canny operator for low-threshold edge detection, extracting edge gradient abrupt change points of the vehicle's physical structure, and widening the edge width through morphological dilation operations.

[0015] As a further improvement of the present invention, the specific process of generating the vehicle body feature contour is as follows: obtaining the moving pixel region, LBP high texture region, weak shadow boundary region, and edge gradient region; selecting the overlapping region that simultaneously covers ≥3 types of features, and using the morphological closing operation of 3×3 rectangular structural elements to fill the pixel holes in the overlapping region to form a vehicle region mask.

[0016] As a further improvement of the present invention, the specific process of generating the vehicle body feature contour further includes: performing contour detection on the vehicle area mask, extracting the maximum circumscribed contour of the vehicle area mask, and then optimizing the contour edge through a polygon approximation algorithm to generate the vehicle body feature contour.

[0017] As a further improvement of the present invention, the specific process of restoring the monitoring video segment with broken trajectory is as follows: extract the vehicle body feature contour of each frame within the video time interval and record the coordinates of the center point of the vehicle body feature contour; obtain the effective trajectory center point coordinates of the last frame of the vehicle body feature contour before the trajectory breakage, and record them as the reference point before the breakage; and obtain the normal trajectory center point coordinates of the first frame of the vehicle body feature contour after the trajectory breakage, and record them as the reference point after the breakage; combine the reference point coordinates before the breakage and the reference point coordinates after the breakage, as well as the coordinates of the center point of the vehicle body feature contour of each frame within the video time interval, to calculate the core motion parameters of the vehicle, including driving speed and driving direction angle; if the vehicle body contour is not generated, the center point coordinates predicted by inertia are completed based on the driving speed and driving direction angle; integrate the center point coordinates of the vehicle body feature contour and the center point coordinates predicted by inertia, and arrange them in the time order of the video frames to form a continuous sequence of trajectory points within the video time interval.

[0018] As a further improvement of the present invention, the specific process of restoring the monitoring video segment with broken trajectory also includes: splicing the continuous trajectory point sequence with the normal trajectory point sequences of all video frames before the break and all video frames after the break to obtain a complete trajectory point sequence, and using a sliding window smoothing algorithm to optimize the complete trajectory point sequence to complete the restoration of the broken trajectory segment.

[0019] This invention also provides an intelligent monitoring and management method for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance, including: Step S10: acquiring vehicle monitoring video and determining whether there is a broken vehicle driving trajectory in the vehicle monitoring video; Step S20: if there is a broken vehicle driving trajectory in the vehicle monitoring video, analyzing whether there is mutual interference between the snow background features and the vehicle body features in the vehicle monitoring video. If there is mutual interference, processing the vehicle monitoring video to enhance the distinguishability between the snow background features and the vehicle body features; Step S30: based on the vehicle monitoring video with enhanced distinguishability between the snow background features and the vehicle body features, fusing the vehicle body contours in the vehicle monitoring video to generate a vehicle body feature contour; Step S40: based on the generated vehicle body feature contour, restoring the broken trajectory monitoring video segments to generate a complete trajectory point sequence for predicting the vehicle's driving intention.

[0020] The beneficial effects of this invention are as follows: 1. In response to the problem of color interference between snow and vehicle body in harsh environments such as snow, the invention effectively solves the problem of vehicle trajectory breakage through multi-feature fusion and enhancement technology, ensuring the accuracy of vehicle body contour extraction and trajectory restoration in complex scenarios, filling the technical gap of trajectory tracking in special weather conditions under traditional monitoring, and accurately judging whether fuel vehicles intend to occupy charging spaces by relying on complete trajectory point sequences and AI intent prediction models, and linking ground locks to achieve automated and intelligent management, avoiding the idle charging piles caused by fuel vehicles occupying spaces from the source, and greatly improving the utilization efficiency and operating income of charging piles.

[0021] 2. The system adopts an automated process of IoT video monitoring, AI algorithm analysis, and equipment linkage throughout the entire process, eliminating the need for manual on-site judgment or handling, thus reducing manpower input and management costs. At the same time, the technical solution has strong compatibility and can be adapted to the transformation of existing charging pile scenarios. It has low construction difficulty and strong implementation. It integrates core technologies such as multi-dimensional feature enhancement, contour reconstruction, and trajectory restoration, which can not only cope with snow scenarios, but also adapt to various complex environments that are prone to background and vehicle interference, improving the system's adaptability to different application scenarios. Through a data-driven decision-making model, the management of charging pile parking spaces becomes more scientific and refined. Attached Figure Description

[0022] The invention will now be further described with reference to the accompanying drawings.

[0023] Figure 1 is a system module diagram of the intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance according to the present invention; Figure 2 is a flowchart of the steps of the intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance according to the present invention. Detailed Implementation

[0024] To make the technical means, creative features, objectives, and effects of this invention easily understood, the application scenario of this invention is based on the premise that the vehicle attributes have been initially identified (e.g., the vehicle has been verified as a fuel vehicle by the identification device at the gate lifting point, hereinafter referred to as the vehicle). The invention will be further described below with reference to specific embodiments. Example 1

[0025] As shown in Figure 1, the intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance according to this embodiment of the invention includes: a trajectory breakage detection module: acquiring vehicle monitoring video and determining whether there is a breakage in the vehicle's driving trajectory in the vehicle monitoring video; the specific process of acquiring vehicle monitoring video in the trajectory breakage detection module is as follows: after the vehicle passes through the barrier gate lifting device and completes the oil-fueled vehicle attribute verification, a continuous video stream of the area where the vehicle is located is collected in real time through a camera, and the video stream is segmented and marked according to the vehicle verification timestamp and camera ID, finally acquiring a complete and continuous vehicle monitoring video from the moment the vehicle enters the monitoring range; the specific process of determining whether there is a breakage in the vehicle's driving trajectory in the trajectory breakage detection module is as follows: The process involves extracting vehicle feature points (including motion pixel cluster centers and high texture entropy points) frame by frame from the acquired vehicle monitoring video. The Hungarian algorithm is then used to match and associate feature points between adjacent frames (for example, the motion pixel cluster center of frame t is matched with the motion pixel cluster center of frame t+1, or the high texture entropy point of frame t is matched with the high texture entropy point of frame t+1, according to the same type of feature points). The feature point matching rate of three consecutive frames is calculated (matching rate = number of matched feature points / total number of feature points). Simultaneously, the number of consecutive frames without effective feature point association is recorded (for example, if no effective feature points based on motion pixel cluster centers and high texture entropy points are matched in frames t+2 and t+3, the number of consecutive unassociated frames is [value missing]. 2) If the feature point matching rate of two or more consecutive frames is lower than a preset threshold (e.g., 20%), and the number of frames without valid feature point association is ≥2, then it is determined that the vehicle monitoring video has a broken vehicle trajectory, and the video time interval corresponding to the broken vehicle trajectory is marked (e.g., from frame t+2 to frame t+5); Interference analysis and feature enhancement module: If the vehicle monitoring video has a broken vehicle trajectory, then it analyzes whether there is mutual interference between the snow background features and the vehicle body features in the vehicle monitoring video. If there is mutual interference, then the vehicle monitoring video is processed to enhance the distinguishability between the snow background features and the vehicle body features; In the interference analysis and feature enhancement module, if the vehicle monitoring video has a broken vehicle trajectory The specific process for analyzing whether there is mutual interference between the snow background features and vehicle features in the vehicle monitoring video is as follows: If there is a break in the vehicle driving trajectory in the vehicle monitoring video, firstly, obtain the video time interval corresponding to the break in the vehicle driving trajectory, extract the vehicle candidate region corresponding to the trajectory before the break in the interval, and at the same time, select the snow background region corresponding to the vehicle candidate region (i.e., the remaining region in the video frame excluding the vehicle candidate region); calculate the RGB color similarity between the vehicle candidate region and the snow background region (the value range is 0-100%, the exemplary calculation process is: take the mean of the R, G, and B channels of all pixels in the vehicle candidate region and the background region, and calculate the similarity of the channel mean using the Euclidean distance formula);Calculate the luminance variance between the vehicle candidate region and the snow background region (range 0-255; exemplary calculation process: calculate the mean luminance of the vehicle candidate region and the background region respectively, and take the absolute difference between the two means); calculate the luminance variance between the vehicle candidate region and the snow background region (range 0-5; exemplary calculation process: calculate the LBP (Local Binary Pattern, LBP) texture entropy value of the vehicle candidate region and the background region respectively, and take the absolute difference between the two entropy values); set the color similarity threshold to 90%, the luminance variance threshold to 15, and the texture entropy difference threshold to 0.5, and compare the calculated color similarity, luminance variance, and luminance variance with the corresponding color similarity threshold, luminance variance threshold, and texture entropy difference threshold respectively; if the color similarity ≥ 90%, the luminance variance ≤ 15, and the texture entropy difference ≤ 0.5, and all three conditions are met simultaneously, it is determined that there is mutual interference between the snow background features and the vehicle body features in the vehicle monitoring video; in the interference analysis and feature enhancement module, if there is mutual interference, the vehicle monitoring video will be... The specific process of frequency processing to enhance the distinction between snow background features and vehicle features is as follows: Motion pixel regions are extracted; three-frame difference operations are performed on vehicle monitoring videos with mutual interference to filter out motion pixel regions (i.e., rigid body regions corresponding to the vehicle) that have displacement in consecutive frames of the vehicle monitoring video; the brightness of the original motion pixel regions is increased by 1.2 times to highlight the motion pixel regions, while the brightness of the static snow background is reduced to 0.8 times the original brightness. The initial distinction between the moving vehicle region and the static snow background is achieved through the brightness region distinction of the motion pixel regions; LBP high-texture regions are extracted, and local binary mode (Local Binary Pattern) is performed frame by frame on the vehicle monitoring video. Binary Pattern (LBP) texture filtering is used to amplify the dense texture details of the vehicle body (such as the texture edges of paint and tires) and suppress the uniform, fluffy texture of the snow. This enhances the recognizability of the vehicle body area. Weak shadow boundary regions are extracted by converting vehicle monitoring video clip frames to the HSV color space and adjusting the gain of the luminance channel (V channel) to 1.3 times the original luminance channel. This amplifies the weak shadow band (5-10 pixels wide) formed by shading at the bottom of the vehicle, constructing an invisible boundary between the vehicle body and the snow. Edge gradient regions are extracted using the Canny operator for low-threshold edge detection (low threshold 50, high threshold 60), extracting abrupt changes in edge gradients of the vehicle's physical structure (such as the outline edges of windows and doors). Morphological dilation (3×3 rectangular structuring elements) is used to widen the edge width, enhancing the gradient difference with the snow background area. The vehicle body contour reconstruction module fuses the vehicle body contours in the monitoring video after enhancing the distinction between the snow background features and vehicle body features, generating a vehicle body feature contour.In the vehicle body contour reconstruction module, based on the enhanced snow background features and vehicle body features, the vehicle body contour in the vehicle monitoring video is fused to generate the vehicle body feature contour. The specific process is as follows: From the enhanced snow background features and vehicle body features, moving pixel regions, LBP high-texture regions, weak shadow boundary regions, and edge gradient regions are obtained; overlapping regions that simultaneously cover ≥3 types of features (i.e., moving pixel regions, LBP high-texture regions, weak shadow boundary regions, and edge gradient regions) are selected, and interference regions corresponding to single features are excluded; morphological closing operations of 3×3 rectangular structural elements are used to fill pixel holes in the overlapping regions to form continuous vehicle regions. Domain masking; contour detection is performed on the vehicle region mask (based on OpenCV's findContours algorithm), extracting the maximum bounding contour of the mask; then, the contour edges are optimized using a polygon approximation algorithm (approximation accuracy set to 1 pixel), finally generating a complete and continuous vehicle body feature contour; Trajectory restoration and intent prediction module: based on the generated vehicle body feature contour, the trajectory fragments of the monitoring video are restored, generating a complete trajectory point sequence for predicting the vehicle's driving intent; in the trajectory restoration and intent prediction module, the specific process of restoring the trajectory fragments of the monitoring video based on the generated vehicle body feature contour is as follows: for the video time interval corresponding to the trajectory fragments of the vehicle monitoring video (e.g., frame t+2 to...), the trajectory fragments are... For frame t+5, extract the vehicle body feature contours for each frame within the corresponding video time interval of the vehicle monitoring video (the vehicle body feature contours are enhanced by the interference analysis and feature enhancement module and generated by the vehicle body contour reconstruction module), and record the center point coordinates of the vehicle body feature contours for each frame (e.g., coordinates (x1, y1) for frame t+2 and (x2, y2) for frame t+3). Obtain the effective trajectory center point coordinates of the vehicle body feature contours in the last frame before trajectory breakage (e.g., frame t+1), and record them as the reference point before breakage. Then, obtain the normal trajectory center point coordinates of the vehicle body feature contours in the first frame after trajectory breakage (e.g., frame t+6), and record them as the reference point after breakage. Combine the coordinates of the reference point before breakage and the reference point after breakage, as well as the trajectory breakage... The coordinates of the center point of the vehicle body feature contour in each frame of the video time interval corresponding to the vehicle monitoring video are used to calculate the core motion parameters of the vehicle, including the driving speed (unit: pixels / second, which can be mapped to actual meters / second) and the driving direction angle (unit: °, i.e. the angle between the trajectory and the horizontal axis). For each frame of the video time interval corresponding to the vehicle monitoring video with broken trajectory (frame t+2 to frame t+5), if extreme interference occurs and the vehicle body contour is not generated, the center point coordinates predicted by inertia are completed based on the calculated driving speed and driving direction angle (example: based on the center point coordinates (x1, y1) of frame t+2 and the driving parameters, the center point coordinates (x2, y2) of frame t+3 are predicted).The center point coordinates of the extracted vehicle body feature contours are integrated with the center point coordinates predicted by inertia and arranged in the video frame time sequence (t+2→t+3→t+4→t+5) to form a continuous trajectory point sequence within the video time interval corresponding to the vehicle monitoring video with trajectory breakage. The completed trajectory point sequence is then concatenated with the normal trajectory point sequences of all video frames before the breakage (e.g., frames 1 to t+1) and all video frames after the breakage (frame t+6 and subsequent) to obtain a complete trajectory point sequence. A sliding window smoothing algorithm (e.g., taking the average coordinates of 3 consecutive trajectory points) is used to optimize the concatenated complete trajectory point sequence, eliminating minor jitter points. Finally, a smooth, continuous, and accurately reconstructed trajectory that matches the actual driving path of the vehicle is obtained, completing the restoration of broken trajectory segments. In the trajectory restoration and intent prediction module, the specific process for predicting the vehicle's driving intent is as follows: Based on the generated complete vehicle trajectory point sequence, the complete vehicle trajectory point sequence is input into an artificial intelligence model with trajectory analysis capabilities (including but not limited to trajectory trend prediction models and path fitting analysis models). The model analyzes the direction and path trend of the vehicle's driving trajectory to predict the vehicle's driving intent. Based on the vehicle's driving intent, the charging pile parking space lock is linked to perform corresponding control actions to prevent gasoline vehicles from occupying charging spaces. Example 2;

[0026] Based on the specific implementation process of Embodiment 1, the present invention provides a smart monitoring and management method for preventing oil vehicle occupancy of charging piles based on Internet of Things video surveillance, including: Step S10: acquiring vehicle monitoring video and determining whether there is a broken vehicle driving trajectory in the vehicle monitoring video; the specific process of acquiring vehicle monitoring video is as follows: acquiring continuous video streams of the area where the vehicle is located in real time through a camera, segmenting and marking the video stream according to the vehicle verification timestamp and camera ID, and acquiring complete vehicle monitoring video.

[0027] The specific process for determining whether there is a vehicle trajectory break in the vehicle monitoring video is as follows: extract motion pixel cluster centers and high texture entropy point feature points from the vehicle monitoring video; use the Hungarian algorithm to match the motion pixel cluster centers and high texture entropy points of the same type between adjacent frames; calculate the feature point matching rate of 3 consecutive frames and record the number of consecutive frames without effective feature point association; if the feature point matching rate of 2 or more consecutive frames is lower than a preset threshold and the number of frames without effective feature point association is ≥2, then it is determined that there is a vehicle trajectory break in the vehicle monitoring video, and the video time interval corresponding to the vehicle trajectory break phenomenon is marked.

[0028] Step S20: If the vehicle monitoring video shows a broken vehicle trajectory, analyze whether there is mutual interference between the snow background features and the vehicle body features in the video. If there is mutual interference, process the video to enhance the distinguishability between the snow background features and the vehicle body features. The specific process for analyzing whether there is mutual interference between the snow background features and the vehicle body features in the video is as follows: extract the vehicle candidate region corresponding to the trajectory break within the video time interval, and select the snow background region corresponding to the vehicle candidate region; calculate the RGB color similarity, brightness variance, and brightness variance of the vehicle candidate region and the snow background region respectively; if the color similarity is ≥90%, the brightness variance is ≤15, and the texture entropy difference is ≤0.5, and all three conditions are met, it is determined that there is mutual interference between the snow background features and the vehicle body features in the video.

[0029] The specific process for enhancing the distinction between snow background features and vehicle body features is as follows: extract moving pixel regions, filter out moving pixel regions that have generated displacement, brighten the original moving pixel regions by 1.2 times, and reduce the brightness of the static snow background to 0.8 times the original brightness; extract LBP high texture regions, perform local binary mode texture filtering on the vehicle monitoring video frame by frame, amplify the dense texture details of the vehicle body, and suppress the uniform fluffy texture of the snow.

[0030] The specific process of enhancing the distinction between snow background features and vehicle body features also includes: extracting weak shadow boundary regions, converting vehicle monitoring video clip frames to HSV color space, adjusting the gain of the brightness channel to 1.3 times the original brightness channel, and amplifying the weak shadow band formed by shading at the bottom of the vehicle; extracting edge gradient regions, using the Canny operator for low-threshold edge detection, extracting edge gradient abrupt change points of the vehicle's physical structure, and widening the edge width through morphological dilation operations.

[0031] Step S30: Based on the enhanced snow background features and vehicle body features, the vehicle body contours in the vehicle monitoring video are fused to generate vehicle body feature contours. The specific process of generating vehicle body feature contours is as follows: obtain the moving pixel region, LBP high texture region, weak shadow boundary region, and edge gradient region; select the overlapping region that covers ≥3 types of features at the same time, and use the morphological closing operation of 3×3 rectangular structural elements to fill the pixel holes in the overlapping region to form a vehicle region mask.

[0032] The specific process of generating the vehicle body feature contour also includes: performing contour detection on the vehicle area mask, extracting the maximum circumscribed contour of the vehicle area mask, and then optimizing the contour edge through a polygon approximation algorithm to generate the vehicle body feature contour.

[0033] Step S40: Based on the generated vehicle body feature contour, restore the broken trajectory monitoring video segment to generate a complete trajectory point sequence for predicting the vehicle's driving intention.

[0034] The specific process for restoring the monitoring video segment with broken trajectory is as follows: extract the vehicle body feature contour of each frame within the video time interval and record the coordinates of the center point of the vehicle body feature contour; obtain the effective trajectory center point coordinates of the last frame of the vehicle body feature contour before the trajectory breakage, and record them as the reference point before the breakage; and obtain the normal trajectory center point coordinates of the first frame of the vehicle body feature contour after the trajectory breakage, and record them as the reference point after the breakage; combine the coordinates of the reference point before the breakage and the reference point after the breakage, as well as the coordinates of the center point of the vehicle body feature contour of each frame within the video time interval, to calculate the core motion parameters of the vehicle, including driving speed and driving direction angle. If the vehicle body contour is not generated, the center point coordinates predicted by inertia are completed based on the driving speed and driving direction angle; integrate the center point coordinates of the vehicle body feature contour and the center point coordinates predicted by inertia, and arrange them in the time order of the video frames to form a continuous sequence of trajectory points within the video time interval.

[0035] The specific process of restoring the broken trajectory monitoring video segment also includes: splicing the continuous trajectory point sequence with the normal trajectory point sequences of all video frames before the break and all video frames after the break to obtain a complete trajectory point sequence, and using a sliding window smoothing algorithm to optimize the complete trajectory point sequence to complete the restoration of the broken trajectory segment.

[0036] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart monitoring and management system for preventing fuel-powered vehicles from occupying charging pile spaces, based on IoT video surveillance, characterized in that: include: Trajectory breakage detection module: acquires vehicle monitoring video and determines whether there is a breakage in the vehicle's driving trajectory in the vehicle monitoring video; Interference Analysis and Feature Enhancement Module: If the vehicle monitoring video shows broken vehicle trajectory, the module analyzes whether there is mutual interference between the snow background features and the vehicle body features in the video. If mutual interference exists, the module processes the video to enhance the distinguishability between the snow background features and the vehicle body features. Vehicle Body Contour Reconstruction Module: Based on the enhanced distinguishability between the snow background features and the vehicle body features, the module fuses the vehicle body contours in the video to generate a vehicle body feature contour. Trajectory Restoration and Intent Prediction Module: Based on the generated vehicle body feature contour, the module restores the broken trajectory video segments to generate a complete trajectory point sequence for predicting the vehicle's driving intent.

2. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 1, characterized in that: The specific process for obtaining vehicle monitoring video is as follows: continuously capture video streams of the area where the vehicle is located in real time through a camera, segment and mark the video streams according to the vehicle verification timestamp and camera ID, and obtain complete vehicle monitoring video.

3. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 1, characterized in that: The specific process for determining whether there is a broken vehicle trajectory in the vehicle monitoring video is as follows: extract the motion pixel cluster center and high texture entropy value feature points from the vehicle monitoring video; use the Hungarian algorithm to match the motion pixel cluster center and high texture entropy value feature points of the same type between adjacent frames; calculate the feature point matching rate of 3 consecutive frames; and record the number of consecutive frames without effective feature point association. If the feature point matching rate of two or more consecutive frames is lower than the preset threshold, and the number of frames with no valid feature point association is ≥2, then it is determined that there is a vehicle trajectory break in the vehicle monitoring video, and the video time interval corresponding to the vehicle trajectory break is marked.

4. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 1, characterized in that: The specific process for analyzing whether there is mutual interference between snow background features and vehicle body features in vehicle monitoring videos is as follows: extract the vehicle candidate region corresponding to the trajectory break within the video time interval, and at the same time select the snow background region corresponding to the vehicle candidate region. Calculate the RGB color similarity, brightness variance, and brightness variance between the vehicle candidate region and the snow background region respectively. If the color similarity is ≥90%, the brightness variance is ≤15, and the texture entropy difference is ≤0.5, and all three conditions are met, then it is determined that there is mutual interference between the snow background features and the vehicle body features in the vehicle monitoring video.

5. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 1, characterized in that: The specific process for enhancing the distinction between snow background features and vehicle body features is as follows: extract moving pixel regions, filter out moving pixel regions that have generated displacement, brighten the original moving pixel regions by 1.2 times, and reduce the brightness of the static snow background to 0.8 times the original brightness; extract LBP high texture regions, perform local binary mode texture filtering on the vehicle monitoring video frame by frame, amplify the dense texture details of the vehicle body, and suppress the uniform fluffy texture of the snow.

6. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 5, characterized in that: The specific process of enhancing the distinction between snow background features and vehicle body features also includes: extracting weak shadow boundary regions, converting vehicle monitoring video clip frames to HSV color space, adjusting the gain of the brightness channel to 1.3 times the original brightness channel, and amplifying the weak shadow band formed by shading at the bottom of the vehicle; extracting edge gradient regions, using the Canny operator for low-threshold edge detection, extracting edge gradient abrupt change points of the vehicle's physical structure, and widening the edge width through morphological dilation operations.

7. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 1, characterized in that: The specific process for generating the vehicle body feature contour is as follows: obtain the moving pixel region, LBP high texture region, weak shadow boundary region, and edge gradient region; select the overlapping region that simultaneously covers ≥3 types of features, and use the morphological closing operation of 3×3 rectangular structural elements to fill the pixel holes in the overlapping region to form a vehicle region mask.

8. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 7, characterized in that: The specific process of generating the vehicle body feature contour also includes: performing contour detection on the vehicle area mask, extracting the maximum circumscribed contour of the vehicle area mask, and then optimizing the contour edge through a polygon approximation algorithm to generate the vehicle body feature contour.

9. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 1, characterized in that: The specific process for restoring the monitoring video segment with broken trajectory is as follows: extract the vehicle body feature contour of each frame within the video time interval and record the coordinates of the center point of the vehicle body feature contour; obtain the effective trajectory center point coordinates of the last frame of the vehicle body feature contour before the trajectory breakage, and record them as the reference point before the breakage; and obtain the normal trajectory center point coordinates of the first frame of the vehicle body feature contour after the trajectory breakage, and record them as the reference point after the breakage. By combining the coordinates of the reference points before and after the breakage, as well as the coordinates of the center point of the vehicle body feature contour in each frame within the video time interval, the core motion parameters of the vehicle, including driving speed and driving direction angle, are calculated. If the vehicle body contour is not generated, the center point coordinates predicted by inertia are completed based on the driving speed and driving direction angle. The coordinates of the center point of the vehicle body feature contour and the center point coordinates predicted by inertia are integrated and arranged in the time sequence of the video frames to form a continuous sequence of trajectory points within the video time interval.

10. The intelligent monitoring and management system for preventing oil-fueled vehicles from occupying charging piles based on IoT video surveillance as described in claim 9, characterized in that: The specific process of restoring the broken trajectory monitoring video segment also includes: splicing the continuous trajectory point sequence with the normal trajectory point sequences of all video frames before the break and all video frames after the break to obtain a complete trajectory point sequence, and using a sliding window smoothing algorithm to optimize the complete trajectory point sequence to complete the restoration of the broken trajectory segment.

Images