Video monitoring method and system of intelligent parking wireless video equipment

Abstract

The invention relates to a video monitoring method and system for intelligent parking wireless video equipment, and the method comprises the following steps: collecting a parking space image of a parking space region, and detecting whether the parking space image has license plate information attached to a vehicle or not; if yes, character and color recognition is carried out on the license plate information, and a license plate color number is obtained; if not, determining a parking space scene state of the parking space area based on the parking space image; and synchronizing to a parking management cloud platform based on the license plate color number or the parking space scene state, so that the management platform performs order charging management and parking space resource scheduling, and the technical problem of difficulty in urban parking is solved.

Description

Technical Field

[0001] This invention relates to the field of image technology, and in particular to a video monitoring method and system for smart parking wireless video equipment. Background Technology

[0002] In smart city transportation systems, roadside parking management is a crucial component. Effective parking management can alleviate urban parking difficulties, improve parking space turnover, and reduce road congestion. Traditional parking management methods, such as magnetic sensors and gate barriers, have drawbacks such as limited functionality, incomplete information collection, and the need for manual intervention.

[0003] With the development of computer vision technology, video-based parking monitoring solutions are gradually becoming mainstream. These solutions deploy camera equipment in parking lots or along roadsides and use image recognition algorithms to automatically detect parking space status and vehicle information, making unmanned and automated parking management possible. Summary of the Invention

[0004] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention proposes a video monitoring method for a smart parking wireless video device, comprising the following steps: Collect parking space images of the parking space area and detect whether the parking space images contain license plate information attached to the vehicle; If the license plate information exists, then character and color recognition is performed on the license plate information to obtain the license plate color number; If it does not exist, the parking scene status of the parking space area is determined based on the parking space image; The information is synchronized to the parking management cloud platform based on the license plate color number or the parking space scene status, so that the management platform can perform order billing management and parking space resource scheduling.

[0005] Furthermore, images of parking spaces in the parking area are collected, including: Initial images of the parking space area are captured from multiple angles by camera terminals deployed on sidewalks or curbs, and the initial images are then stitched together to obtain a panoramic parking space image. The panoramic parking space image is clearly optimized, and the parking space boundary lines in the clearly optimized panoramic parking space image are identified and marked. At the same time, the clearly optimized panoramic parking space image is divided into regions based on the parking space boundary lines to obtain parking space images; wherein, the parking space image corresponds to a parking space region.

[0006] Furthermore, detecting whether the parking space image contains license plate information associated with the vehicle includes: The parking space image is input into the YOLOv5 model; wherein the YOLOv5 model includes a Focus slicing module, a CSPDarknet53 backbone feature extraction module, an SPP spatial pyramid pooling module, and a YOLOHead detection output module. The parking space image is sliced ​​using the Focus slicing module to obtain a preliminary slice feature map. The CSPDarknet53 backbone feature extraction module is used to extract depth features from the preliminary slice feature map to obtain a multi-scale depth feature map. The SPP spatial pyramid pooling module is used to perform spatial information fusion on the multi-scale depth feature map to obtain a pooled fused feature map. The pooled fusion feature map is input into the YOLOHead detection output module to determine whether the parking space image contains a license plate attached to a vehicle.

[0007] Furthermore, character and color recognition is performed on the license plate information to obtain the license plate color number, including: The license plate information is input into a preset ResNet model; wherein the ResNet model includes a convolutional feature extraction module, a residual bottleneck module, and a fully connected classification module; The license plate information is input into the convolutional feature extraction module for multi-scale feature extraction and edge contour detection to obtain a primary feature map of the license plate region. The primary feature map of the license plate region is input into the residual bottleneck module for residual learning and feature optimization operations to obtain a deep semantic feature vector. The deep semantic feature vector is input into the fully connected classification module for probability distribution calculation and category mapping to obtain the license plate color number.

[0008] Further, after obtaining the license plate color number, the steps include: Based on the license plate color number, the system clock of the camera terminal is timestamped to obtain the vehicle entry timestamp, and the vehicle entry timestamp is then converted to a standard time format. By associating and binding the standard time format with the license plate color number, a vehicle parking start record is obtained, and the time information of the vehicle entering the corresponding parking space area is determined based on the vehicle parking start record.

[0009] Furthermore, determining the parking scene state of the parking space area based on the parking space image includes: The pixel grayscale values ​​of the parking space image are statistically analyzed to obtain the overall grayscale distribution map of the parking space area. The overall grayscale distribution map is then segmented by grayscale thresholding to obtain a grayscale segmented image. The grayscale segmented image is subjected to morphological processing to obtain a morphologically processed image, and connected component analysis is performed on the morphologically processed image to determine the area and shape features of each connected component. The scene state of the parking space is determined based on its area and shape features.

[0010] Furthermore, based on area and shape features, the scene state of the parking space area is determined to obtain the parking space scene state, including: The area of ​​each connected component is compared with a threshold to obtain the area comparison result, and the shape features of each connected component are analyzed for contour regularity to obtain the shape regularity analysis result. Based on the area comparison results and the shape regularity analysis results, the scene state of the parking space area is determined to obtain the parking space scene state.

[0011] The present invention also provides a video monitoring system for a smart parking wireless video device, comprising: The acquisition module is used to acquire images of parking spaces in the parking area and detect whether the parking space images contain license plate information attached to vehicles. The recognition module is used to perform character and color recognition on the license plate information when it exists, and obtain the license plate color number; The determination module is used to determine the parking scene status of the parking space area based on the parking space image when the parking space does not exist. The billing module is used to synchronize the license plate color number or the parking space scene status to the parking management cloud platform, so that the management platform can perform order billing management and parking space resource scheduling.

[0012] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the above methods.

[0013] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the above methods.

[0014] This invention provides a video monitoring method for a smart parking wireless video device, comprising the following steps: acquiring parking space images of a parking area and detecting whether the parking space images contain license plate information attached to a vehicle; if they do, performing character and color recognition on the license plate information to obtain a license plate color number; if they do not, determining the parking space scene status of the parking area based on the parking space images; and synchronizing the information to a parking management cloud platform based on the license plate color number or the parking space scene status for the management platform to perform order billing management and parking space resource scheduling. This solves the technical problem of urban parking difficulties. By using wireless video devices as front-end sensing units, it eliminates the problem of cumbersome wiring. At the same time, it reduces invalid data uploads through local intelligent judgment, saving communication costs and extending device battery life. It is suitable for rapid deployment and long-term stable operation in large-scale distributed parking scenarios. Attached Figure Description

[0015] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a flowchart illustrating the steps of the video monitoring method for a smart parking wireless video device in an embodiment of the present invention. Figure 2 This is a structural block diagram of the video monitoring system of the smart parking wireless video device in this embodiment of the invention; Figure 3 This is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0016] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0018] In the description of this invention, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0019] In the description of this invention, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0020] In the description of this invention, unless otherwise explicitly defined, terms such as "set up," "install," and "connect" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0021] The embodiments of this application will be further described below with reference to the accompanying drawings.

[0022] Reference Figure 1 This invention provides a video monitoring method for a smart parking wireless video device, comprising the following steps: Step S1: Collect parking space images of the parking space area and detect whether the parking space images contain license plate information associated with the vehicle.

[0023] Specifically, cameras deployed around the parking spaces capture complete images of the parking area, ensuring the lenses cover parking space markings and areas where license plates may appear. Exposure parameters are adjusted to suit ambient lighting conditions during capture to avoid image blurring caused by strong light or shadows. The acquired parking space images are then imported into a pre-defined detection system. Image preprocessing removes noise and enhances contrast. A license plate feature recognition algorithm scans areas in the image that conform to rectangular outlines and character arrangement patterns, such as identifying the color and character distribution characteristics of blue-background, white-letter license plates. By comparing this information with a license plate template library, it determines whether the parking space image contains the license plate information of a vehicle.

[0024] Step S2: If the license plate information exists, perform character and color recognition to obtain the license plate color number.

[0025] Specifically, when the parking space image contains license plate information, the license plate area is first precisely cropped. Color channels are separated through color space conversion, and the peak pixel distribution of each channel is analyzed to determine if the license plate background color is blue, yellow, etc. For example, a blue license plate will have a significantly higher value in channel B than in other channels. Next, character segmentation is performed on the cropped license plate image to remove borders and rivet interference. Through character contour comparison and stroke feature matching, Chinese characters, letters, and numbers are identified. For example, the stroke transition differences between "Z" and "2" are distinguished. Finally, the color determination results and character recognition results are integrated to obtain the complete license plate color number.

[0026] Step S3: If no such condition exists, determine the parking scene status of the parking space area based on the parking space image.

[0027] Specifically, when the parking space image does not contain license plate information, the image is first converted to grayscale, and the pixel brightness distribution within the parking space area is statistically analyzed. A grayscale threshold is set to distinguish the foreground from the background. Next, contour extraction is performed on the processed image to analyze whether there are large, continuous contours within the parking space, such as the rectangular contours of a sedan or the large, irregular contours of an SUV. The positional relationship between the contours and the parking space markings is also compared. If the contour occupies more than 60% of the total area of ​​the parking space and its edges overlap with the markings, it is determined to be occupied by a vehicle. If only the complete parking space markings can be identified, and there are no obvious large contours, it is determined to be vacant, thus determining the parking space scene status.

[0028] Step S4: Based on the license plate color number or the parking space scene status, synchronize to the parking management cloud platform so that the management platform can perform order billing management and parking space resource scheduling.

[0029] Specifically, the acquired license plate color number or parking space status is first standardized according to the data format preset by the cloud platform. For example, the license plate color number is labeled as "Blue-Guangdong A12345", and the parking space status is labeled as "Idle" or "Occupied". The standardized data is then synchronized to the parking management cloud platform in real time through an encrypted network transmission channel. After receiving the data, the platform automatically associates it with the basic information of the corresponding parking space. If the synchronized data is the license plate color number, the platform retrieves the entry time associated with it to start billing; if the synchronized data is the parking space status, the platform updates the parking space resource distribution map, for example, marking "Idle" parking spaces as available for reservation, providing data support for order billing management and parking space resource scheduling.

[0030] In a specific embodiment, acquiring parking space images of the parking space area includes: Initial images of the parking space area are captured from multiple angles by camera terminals deployed on sidewalks or curbs, and the initial images are then stitched together to obtain a panoramic parking space image. The panoramic parking space image is clearly optimized, and the parking space boundary lines in the clearly optimized panoramic parking space image are identified and marked. At the same time, the clearly optimized panoramic parking space image is divided into regions based on the parking space boundary lines to obtain parking space images; wherein, the parking space image corresponds to a parking space region.

[0031] Specifically, camera terminals are fixedly installed at suitable locations on sidewalks or curbs. The installation height and angle of the terminals are adjusted according to the distribution of parking spaces to ensure that each camera terminal can cover multiple adjacent parking spaces, and that the shooting angles of different terminals partially overlap. For example, one camera is placed every 5 meters along both sides of the parking lot aisle, with the lens tilted at about 30 degrees towards the parking space area, to capture initial images of the parking spaces and their surrounding environment from multiple angles. Then, an image stitching algorithm is used to extract overlapping feature points from each initial image, such as intersections of parking space markings and corners of the curb, and the multiple initial images are seamlessly merged to form a panoramic parking space image that completely covers the target area. Next, the panoramic parking space image is subjected to noise reduction and sharpening to enhance the contrast of the parking space boundary lines in the image. Then, edge detection technology is used to identify and mark the white or yellow parking space boundary lines. Based on the closed contours of the marked parking space boundary lines, the clear and optimized panoramic parking space image is divided into multiple independent regions. Each closed contour corresponds to a parking space region, and finally, a parking space image that accurately corresponds to a single parking space region is obtained. For example, if 8 closed boundary lines are identified in a panoramic image, 8 parking space images corresponding to different parking space regions are obtained.

[0032] In a specific embodiment, detecting whether the parking space image contains license plate information associated with the vehicle includes: The parking space image is input into the YOLOv5 model; wherein the YOLOv5 model includes a Focus slicing module, a CSPDarknet53 backbone feature extraction module, an SPP spatial pyramid pooling module, and a YOLOHead detection output module. The parking space image is sliced ​​using the Focus slicing module to obtain a preliminary slice feature map. The CSPDarknet53 backbone feature extraction module is used to extract depth features from the preliminary slice feature map to obtain a multi-scale depth feature map. The SPP spatial pyramid pooling module is used to perform spatial information fusion on the multi-scale depth feature map to obtain a pooled fused feature map. The pooled fusion feature map is input into the YOLOHead detection output module to determine whether the parking space image contains a license plate attached to a vehicle.

[0033] Specifically, the parking space image corresponding to a single parking space area is standardized according to the input format required by the YOLOv5 model and then directly input into the model. The model's built-in Focus slicing module, CSPDarknet53 backbone feature extraction module, SPP spatial pyramid pooling module, and YOLOHead detection output module will then function sequentially. First, the Focus slicing module performs a block slicing operation on the parking space image, dividing the original image into multiple sub-image blocks at fixed pixel intervals. These sub-blocks are then stacked and combined to form a preliminary slice feature map. For example, a 640×640 pixel parking space image is sliced ​​and transformed into a 320×320 pixel feature map. Next, the preliminary slice feature map is fed into the CSPDarknet53 backbone feature extraction module. Through repeated convolution operations using multiple residual network blocks, deep features such as texture and contours in the image are gradually extracted. At the same time, feature maps of different scales are generated to adapt to the needs of detecting license plate targets of different sizes. Subsequently, the multi-scale deep feature maps are fed into the SPP spatial pyramid pooling module. Spatial information is sampled and fused using pooling kernels of different sizes, such as 13×13, 9×9, and 5×5 kernels. This integrates feature information from different receptive fields into a unified-dimensional pooled fusion feature map, enhancing the ability to represent license plate targets. Finally, the pooled fusion feature map is input into the YOLOHead detection output module. This module uses anchor box matching and classification regression to identify targets in the feature map. If a target region matching license plate characteristics is detected and the confidence level exceeds a preset threshold, it is determined that the parking space image contains license plate information associated with a vehicle; otherwise, it is determined that it does not.

[0034] In a specific embodiment, character and color recognition is performed on the license plate information to obtain the license plate color number, including: The license plate information is input into a preset ResNet model; wherein the ResNet model includes a convolutional feature extraction module, a residual bottleneck module, and a fully connected classification module; The license plate information is input into the convolutional feature extraction module for multi-scale feature extraction and edge contour detection to obtain a primary feature map of the license plate region. The primary feature map of the license plate region is input into the residual bottleneck module for residual learning and feature optimization operations to obtain a deep semantic feature vector. The deep semantic feature vector is input into the fully connected classification module for probability distribution calculation and category mapping to obtain the license plate color number.

[0035] Specifically, first adjust the previously detected license plate information to a preset format to ensure that it can meet the input requirements of the ResNet model, and then input it into the model. The convolutional feature extraction module, residual bottleneck module, and fully connected classification module in the model will process this information in sequence. First, the license plate information enters the convolutional feature extraction module. Convolution kernels of different sizes in the module will scan the license plate information layer by layer. For example, a small 3×3 convolutional kernel is used to capture the edge details of license plate characters, and a 5×5 convolutional kernel is used to extract the overall contour features of the license plate. At the same time, combined with an edge detection algorithm, the contrast between characters and the background is enhanced, so that a primary feature map of the license plate area containing basic visual information of the license plate can be obtained. Features such as the stroke direction of characters on the license plate and color transition areas will be initially recorded in this step. Then, the primary feature map of the license plate area is passed into the residual bottleneck module. This module skips some convolutional layers through a shortcut connection to avoid feature degradation during deep-layer transmission. For example, when processing complex license plate characters (such as easily confused characters like "6" and "0"), residual learning can retain the subtle difference features of the characters. After multiple feature optimization operations, the primary feature map is transformed into a more recognizable deep semantic feature vector, enabling the vector to accurately represent the character and color information of the license plate. Finally, the deep semantic feature vector is input into the fully connected classification module. The module first calculates the probability distribution of the vector, calculates the probability values corresponding to each possible character and color category. For example, the probability of the "blue background" color category is calculated as 98%, and the probability of the "A" character category is calculated as 95%. Then, the category combination with the highest probability is obtained through category mapping, and finally the complete license plate color number is obtained, such as "blue background Beijing A88888".

[0036] In a specific embodiment, after the step of obtaining the license plate color number, it includes: Obtain a timestamp for the system clock of the camera terminal based on the license plate color number to obtain a vehicle entry timestamp, and perform format conversion processing on the vehicle entry timestamp to obtain a standard time format; Associate and bind the standard time format through the license plate color number to obtain a vehicle parking start record, and determine the time information when the vehicle drives into the corresponding parking space area based on the vehicle parking start record.

[0037] Specifically, first take the obtained license plate color number and retrieve the system clock data from the camera terminal that captured the license plate information. Trigger the time recording function in the terminal to obtain the vehicle entry timestamp, which is accurate to the second level. For example, when a blue license plate "粤B12345" is captured by the camera terminal, the timestamp displayed by the terminal system clock is "1732658923". Then perform format conversion on this vehicle entry timestamp and adjust the format according to the general rule of "year - month - day hour: minute: second". For example, convert "1732658923" to "2024-11-27 15:48:43", thus obtaining the standard time format. After that, bind the license plate color number and the converted standard time format through a data association tool to ensure that the two form a one-to-one correspondence relationship. For example, bind "blue - 粤B12345" and "2024-11-27 15:48:43" together to generate a vehicle parking start record containing license plate information and the corresponding time. Finally, extract the bound standard time format from this vehicle parking start record, and this time is the time information when the vehicle enters the corresponding parking space area. For example, extract "2024-11-27 15:48:43" from the above record as the specific time basis for the blue 粤B12345 vehicle to enter its corresponding parking space area.

[0038] In a specific embodiment, determining the parking space scene state of the parking space area based on the parking space image includes: Perform pixel gray value statistics on the parking space image to obtain the overall gray distribution map of the parking space area, and perform gray threshold segmentation on the overall gray distribution map to obtain a gray segmentation image; Perform morphological processing on the gray segmentation image to obtain an image after morphological processing, and perform connected component analysis on the image after morphological processing to determine the area and shape characteristics of each connected component; Based on the area and shape characteristics, perform scene state judgment on the parking space area to obtain the parking space scene state.

[0039] Specifically, firstly, for parking space images without license plate information, the grayscale value of each pixel is counted row by row and column by column. For example, the grayscale value of ground pixels is mostly between 150-200, while the grayscale value of the vehicle body area may be as low as 50-100 if there is a vehicle. Based on this, an overall grayscale distribution map covering the entire parking space area is generated. Next, an appropriate grayscale threshold is set, such as 120 as the segmentation point. Pixels in the overall grayscale distribution map that are above the threshold are grouped into one category, and those that are below the threshold are grouped into another category, thus obtaining a grayscale segmentation image containing only black and white. When there is no car in the parking space, the grayscale segmentation image is mostly white areas, while when there is a car, obvious black blocks will appear. Subsequently, morphological processing of dilation and erosion is performed on the grayscale segmentation image. The dilation operation can fill the small gaps in the black blocks, such as the breaks caused by vehicle shadows, while the erosion operation can eliminate sporadic black noise in the white areas, such as small spots formed by dirt on the ground, and finally obtain an image with clearer contours after morphological processing. Next, connected component analysis is performed on this image. By scanning the image, the set of interconnected pixels is identified, and the area of ​​each connected component is calculated based on the number of pixels in each component. At the same time, edge fitting is used to determine whether the connected components have shape features such as rectangles or irregular shapes. For example, when there is a car in the parking space, a large connected component with an area close to a rectangle will appear. Finally, the parking space scene status is determined based on preset area and shape standards. For example, when the area of ​​the largest connected component exceeds 60% of the total area of ​​the parking space and the shape is close to a rectangle, it is determined to be in a "car occupied" state. If the area of ​​all connected components is less than 10% of the total area of ​​the parking space and there are large areas with irregular shapes, and only small connected components corresponding to the parking space markings can be identified, it is determined to be in a "no car occupancy" state. This is how the accurate parking space scene status is obtained.

[0040] In a specific embodiment, the scene state of the parking space area is determined based on area and shape features to obtain the parking space scene state, including: The area of ​​each connected component is compared with a threshold to obtain the area comparison result, and the shape features of each connected component are analyzed for contour regularity to obtain the shape regularity analysis result. Based on the area comparison results and the shape regularity analysis results, the scene state of the parking space area is determined to obtain the parking space scene state.

[0041] Specifically, the area data of each connected component obtained earlier is first compared with a preset area threshold range. This threshold is set according to the actual size of the parking space. For example, if the total area of ​​a parking space is 25 square meters, "greater than 15 square meters" is set as the occupancy threshold and "less than 3 square meters" is set as the miscellaneous object threshold. For example, a connected component area of ​​20 square meters meets the occupancy standard, while 1 square meter is classified as miscellaneous object. This yields the area comparison result for each connected component. Next, contour regularity analysis is performed on the shape characteristics of each connected component. Regularity is determined by calculating the concavity and convexity of the contour and the distribution of the number of edges. For example, the contour of a connected component corresponding to a car body is close to a rectangle, with few concave and convex points and a regular number of edges, so the regularity will be higher than 80%. On the other hand, the contour of a connected component formed by pebbles on the ground is irregular and has many concave and convex points, so the regularity may be lower than 30%. This yields the shape regularity analysis result for each connected component.

[0042] Then, the parking space scene status is comprehensively judged by combining the area comparison results and the shape regularity analysis results. For example, if the area comparison result of a certain connected component meets the occupancy threshold and the shape regularity analysis result shows that the regularity is higher than 75%, it is determined that there is a vehicle parked in the parking space area, which is a "car occupied" state. If the area comparison results of all connected components are within the clutter threshold, regardless of the shape regularity, it is determined to be a "no car but clutter" state. If no connected component reaches the occupancy or clutter threshold, and the shape regularity analysis result shows that there are only small regular outlines corresponding to the parking space markings, it is determined to be a "no car vacant" state. Finally, the accurate parking space scene status is obtained through such comprehensive judgment.

[0043] The video monitoring method of the smart parking wireless video device in the embodiments of the present invention has been described above. The video monitoring system of the smart parking wireless video device in the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 2 One embodiment of the video monitoring system for smart parking wireless video equipment in this invention includes: The acquisition module 21 is used to acquire parking space images in the parking space area and detect whether the parking space images contain license plate information attached to the vehicle. The recognition module 22 is used to perform character and color recognition on the license plate information when it exists, and obtain the license plate color number; The determination module 23 is used to determine the parking scene status of the parking space area based on the parking space image when the parking space does not exist. The billing module 24 is used to synchronize the license plate color number or the parking space scene status to the parking management cloud platform so that the management platform can perform order billing management and parking space resource scheduling.

[0044] In this embodiment, the specific implementation of each unit in the above system embodiment is the same as that in the above method embodiment, and will not be repeated here.

[0045] like Figure 3 As shown in the diagram, this embodiment of the invention provides a structural schematic block diagram of a computer device, including: At least one processor; At least one memory for storing at least one program; When at least one program is executed by at least one processor, the at least one processor implements the video monitoring method of the smart parking wireless video device described above.

[0046] It is evident that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented in this device embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0047] Furthermore, this application also discloses a computer program product or computer program stored in a computer-readable storage medium. A processor of a computer device can read the computer program from the computer-readable storage medium and execute the computer program, causing the computer device to perform the video monitoring method of the aforementioned smart parking wireless video device. Similarly, the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0048] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A video monitoring method for a smart parking wireless video device, characterized in that, Includes the following steps: Collect parking space images of the parking space area and detect whether the parking space images contain license plate information attached to the vehicle; If the license plate information exists, then character and color recognition is performed on the license plate information to obtain the license plate color number; If it does not exist, the parking scene status of the parking space area is determined based on the parking space image; The information is synchronized to the parking management cloud platform based on the license plate color number or the parking space scene status, so that the management platform can perform order billing management and parking space resource scheduling.

2. The video monitoring method for the intelligent parking wireless video equipment according to claim 1, characterized in that, Acquire parking space images of the parking area, including: Initial images of the parking space area are captured from multiple angles by camera terminals deployed on sidewalks or curbs, and the initial images are then stitched together to obtain a panoramic parking space image. The panoramic parking space image is clearly optimized, and the parking space boundary lines in the clearly optimized panoramic parking space image are identified and marked. At the same time, the clearly optimized panoramic parking space image is divided into regions based on the parking space boundary lines to obtain parking space images; wherein, the parking space image corresponds to a parking space region.

3. The video monitoring method for the intelligent parking wireless video equipment according to claim 1, characterized in that, Detecting whether the parking space image contains license plate information associated with the vehicle, including: The parking space image is input into the YOLOv5 model; wherein the YOLOv5 model includes a Focus slicing module, a CSPDarknet53 backbone feature extraction module, an SPP spatial pyramid pooling module, and a YOLOHead detection output module. The parking space image is sliced ​​using the Focus slicing module to obtain a preliminary slice feature map. The CSPDarknet53 backbone feature extraction module is used to extract depth features from the preliminary slice feature map to obtain a multi-scale depth feature map. The SPP spatial pyramid pooling module is used to perform spatial information fusion on the multi-scale depth feature map to obtain a pooled fused feature map. The pooled fusion feature map is input into the YOLOHead detection output module to determine whether the parking space image contains a license plate attached to a vehicle.

4. The video monitoring method for the intelligent parking wireless video equipment according to claim 1, characterized in that, The license plate information is subjected to character and color recognition to obtain the license plate color number, including: The license plate information is input into a preset ResNet model; wherein the ResNet model includes a convolutional feature extraction module, a residual bottleneck module, and a fully connected classification module; The license plate information is input into the convolutional feature extraction module for multi-scale feature extraction and edge contour detection to obtain a primary feature map of the license plate region. The primary feature map of the license plate region is input into the residual bottleneck module for residual learning and feature optimization operations to obtain a deep semantic feature vector. The deep semantic feature vector is input into the fully connected classification module for probability distribution calculation and category mapping to obtain the license plate color number.

5. The video monitoring method for the intelligent parking wireless video equipment according to claim 1, characterized in that, After obtaining the license plate color number, the steps include: Based on the license plate color number, the system clock of the camera terminal is timestamped to obtain the vehicle entry timestamp, and the vehicle entry timestamp is then converted to a standard time format. By associating and binding the standard time format with the license plate color number, a vehicle parking start record is obtained, and the time information of the vehicle entering the corresponding parking space area is determined based on the vehicle parking start record.

6. The video monitoring method for the intelligent parking wireless video equipment according to claim 1, characterized in that, Determining the parking scene status of the parking space area based on the parking space image includes: The pixel grayscale values ​​of the parking space image are statistically analyzed to obtain the overall grayscale distribution map of the parking space area. The overall grayscale distribution map is then segmented by grayscale thresholding to obtain a grayscale segmented image. The grayscale segmented image is subjected to morphological processing to obtain a morphologically processed image, and connected component analysis is performed on the morphologically processed image to determine the area and shape features of each connected component. The scene state of the parking space is determined based on its area and shape features.

7. The video monitoring method for the intelligent parking wireless video equipment according to claim 6, characterized in that, The scene state of the parking space area is determined based on its area and shape features, resulting in the parking space scene state, including: The area of ​​each connected component is compared with a threshold to obtain the area comparison result, and the shape features of each connected component are analyzed for contour regularity to obtain the shape regularity analysis result. Based on the area comparison results and the shape regularity analysis results, the scene state of the parking space area is determined to obtain the parking space scene state.

8. A video monitoring system for a smart parking wireless video device, characterized in that, include: The acquisition module is used to acquire images of parking spaces in the parking area and detect whether the parking space images contain license plate information attached to vehicles. The recognition module is used to perform character and color recognition on the license plate information when it exists, and obtain the license plate color number; The determination module is used to determine the parking scene status of the parking space area based on the parking space image when the parking space does not exist. The billing module is used to synchronize the license plate color number or the parking space scene status to the parking management cloud platform, so that the management platform can perform order billing management and parking space resource scheduling.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, The processor executes the steps of any one of claims 1 to 7 when executing a computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When a computer program is executed by a processor, it implements the steps of the method of any one of claims 1 to 7.

Images