Camera management method and monitoring system

Abstract

The application relates to a camera management method and a monitoring system. The method comprises the following steps: acquiring scene information of a target vehicle when it is monitored that a first camera device captures the situation that the target vehicle drives in or drives out; acquiring parking time of the target vehicle if the scene information meets preset scheduling conditions; instructing a second camera device to capture vehicle images of the target vehicle at a preset frequency if the parking time is greater than a preset parking threshold; performing quality evaluation on each vehicle image captured by the second camera device, determining a comprehensive confidence of each vehicle image, and adjusting the capturing frequency of the second camera device according to the comprehensive confidence of each vehicle image, and instructing the second camera device to capture the vehicle images of the target vehicle at the adjusted capturing frequency. The method can effectively manage each camera device in the monitoring system and prolong the service life of the device.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and in particular to a camera management method and monitoring system. Background Technology

[0002] In the field of intelligent transportation technology, cameras are typically used to capture vehicle information, and vehicles are managed based on the captured license plate information, such as illegal parking and vehicle charging.

[0003] In related technologies, when capturing vehicle information, vehicle monitoring is usually carried out based on the linkage of dual cameras. For example, a fixed monitoring camera (bullet camera) is used to monitor the vehicle entering or leaving, and when the vehicle enters or leaves, an integrated spherical camera (PTZ camera) is triggered to rotate to a preset position to capture the license plate, so as to avoid the license plate being obscured.

[0004] However, the timing of camera triggering in related technologies is unreasonable, resulting in excessively long working time for the PTZ camera and shortening its lifespan. Summary of the Invention

[0005] Therefore, it is necessary to provide a camera management method and monitoring system that can effectively manage various camera devices in a monitoring system and extend the service life of the devices, in order to address the above-mentioned technical problems.

[0006] In a first aspect, this application provides a camera management method applied to a controller in a monitoring system, the monitoring system including a first camera device and a second camera device connected to the controller; the method includes:

[0007] If the first camera device captures the target vehicle entering or exiting, obtain scene information of the target vehicle;

[0008] If the scene information meets the preset scheduling conditions, then obtain the stop time of the target vehicle;

[0009] If the stopping time exceeds the preset stopping threshold, the second camera device is instructed to capture vehicle images of the target vehicle at a preset frequency;

[0010] The quality of each vehicle image captured by the second camera is evaluated to determine the overall confidence level of each vehicle image. Based on the overall confidence level of each vehicle image, the capture frequency of the second camera is adjusted, and the second camera is instructed to capture vehicle images of the target vehicle at the adjusted capture frequency.

[0011] In one embodiment, obtaining scene information of the target vehicle includes:

[0012] Acquire vehicle images captured by the first camera and historical video from the second camera;

[0013] The system calls a preset image classification and recognition model to analyze the vehicle images captured by the first camera device to obtain environmental information; and evaluates the similarity and brightness fluctuation of historical videos to determine the device status of the second camera device.

[0014] Environmental information and the device status of the second camera are used as scene information.

[0015] In one embodiment, the similarity and brightness fluctuation of historical videos are evaluated to determine the device status of the second camera device, including:

[0016] A preset number of adjacent frames are obtained from historical videos, and if the structural similarity between each adjacent frame is greater than the similarity threshold, the device status of the second camera is determined to be a stuttering state.

[0017] Calculate the brightness fluctuation value in each adjacent frame, and determine the device state of the second camera device as flickering state if the brightness fluctuation value is greater than the preset fluctuation threshold; the brightness fluctuation value includes the maximum brightness fluctuation amplitude value and the maximum brightness fluctuation frequency.

[0018] In one embodiment, a quality assessment is performed on each vehicle image to determine the overall confidence level of each vehicle image, including:

[0019] For any vehicle image, determine the quantized value of the vehicle image's size features based on the vehicle image's size feature information, and determine the quantized value of the vehicle image's text features based on the vehicle image's license plate information.

[0020] The overall confidence level of the vehicle image is obtained by weighted fusion of the quantized values ​​of size features and text features.

[0021] In one embodiment, the license plate information includes a license plate string; based on the license plate information of the vehicle image, determining the text feature quantization value of the vehicle image includes:

[0022] Determine the character at a preset position from the license plate string;

[0023] The character at the preset position in the license plate string is matched with each character in the first character library to determine the quantization value of a single character;

[0024] The quantization value of a single character is determined as the quantization value of the text feature.

[0025] In one embodiment, the license plate information includes a license plate string; based on the license plate information of the vehicle image, determining the text feature quantization value of the vehicle image includes:

[0026] Each character in the license plate string is matched with each character in the second character library to determine the credibility of each character in the license plate string;

[0027] By comparing the credibility of each character in the license plate string, the minimum credibility is determined as the obfuscated character quantization value;

[0028] The quantization value of the obfuscated character is determined as the quantization value of the text feature.

[0029] In one embodiment, each character in the license plate string is matched with characters in a second character library to determine the credibility of each character in the license plate string, including:

[0030] For each character in the license plate string, calculate the similarity between the character and each character in the second character library;

[0031] Mapping the maximum similarity and the second maximum similarity yields the credibility of each character.

[0032] In one embodiment, the license plate information includes the license plate length and license plate format; based on the license plate information of the vehicle image, the text feature quantization value of the vehicle image is determined, including:

[0033] Color recognition is performed on the license plate information of the vehicle image to determine the vehicle type of the target vehicle;

[0034] Based on the vehicle type of the target vehicle, determine the standard length and standard format of the license plate;

[0035] The standard license plate length is compared with the actual license plate length to determine the license plate length feature value of the vehicle image; and the standard license plate format is matched with the actual license plate format to determine the license plate format feature value of the vehicle image.

[0036] Based on the license plate length feature value and the license plate format feature value, the text feature quantization value of the vehicle image is determined.

[0037] In one embodiment, the capture frequency of the second camera is adjusted based on the overall confidence level of each vehicle image, including:

[0038] If the overall confidence level of all vehicle images is greater than the upper limit of the preset confidence level interval, instruct the second camera to stop capturing images;

[0039] When the overall confidence level of each vehicle image is within the preset confidence level range, the second camera device is instructed to capture images according to the tiered scheduling frequency.

[0040] If the overall confidence level of all vehicle images is less than the lower limit of the preset confidence interval, the second camera device is instructed to stop capturing images.

[0041] In one embodiment, the method further includes:

[0042] If the stopping time is less than a preset stopping threshold, the second camera device is instructed to capture an image of the target vehicle.

[0043] Secondly, this application also provides a monitoring system, which includes: a controller, a first camera device and a second camera device connected to the controller;

[0044] A controller for performing the steps of the method in any of the embodiments of the first aspect described above.

[0045] Thirdly, this application also provides a camera management device, comprising:

[0046] The scene information acquisition module is used to acquire scene information of the target vehicle when the first camera device captures the target vehicle entering or leaving the vehicle.

[0047] The stop time acquisition module is used to acquire the stop time of the target vehicle if the scenario information meets the preset scheduling conditions.

[0048] The stopping time comparison module is used to instruct the second camera device to capture vehicle images of the target vehicle at a preset frequency when the stopping time exceeds a preset stopping threshold.

[0049] The capture frequency adjustment module is used to evaluate the quality of each vehicle image captured by the second camera, determine the overall confidence level of each vehicle image, and adjust the capture frequency of the second camera based on the overall confidence level of each vehicle image, instructing the second camera to capture vehicle images of the target vehicle at the adjusted capture frequency.

[0050] Fourthly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program and the processor executes the steps of the method of any of the embodiments in the first aspect described above.

[0051] Fifthly, this application also provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor of the steps of the method of any of the embodiments of the first aspect described above.

[0052] In a sixth aspect, this application also provides a computer program product, including a computer program that is processed by a processor to perform the steps of the method of any of the embodiments in the first aspect described above.

[0053] The aforementioned camera management method and monitoring system, upon detecting a target vehicle entering or exiting, acquires scene information of the target vehicle. If the scene information meets preset scheduling conditions, the system acquires the target vehicle's stopping time. If the stopping time exceeds a preset stopping threshold, the system instructs a second camera to capture vehicle images of the target vehicle at a preset frequency. The system then evaluates the quality of each captured vehicle image, determines the overall confidence level of each image, and adjusts the capture frequency of the second camera based on the overall confidence level, instructing it to capture vehicle images of the target vehicle at the adjusted frequency. In this method, upon detecting a target vehicle entering or exiting, the system further assesses the scene information and vehicle stopping time. If both meet the scheduling conditions, the system triggers the second camera to start capturing images, reducing the number of times the second camera is activated, thereby reducing its failure rate and extending its lifespan. Next, during the capture process of the second camera, the capture frequency of the second camera is adjusted by evaluating the quality of the captured images to avoid invalid captures, further shorten the working time of the second camera, and extend its service life. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is an application environment diagram of a camera management method in one embodiment;

[0056] Figure 2 This is a flowchart illustrating a camera management method in one embodiment;

[0057] Figure 3 This is a flowchart illustrating the camera management method in another embodiment;

[0058] Figure 4 This is a flowchart illustrating the camera management method in another embodiment;

[0059] Figure 5 This is a flowchart illustrating the camera management method in another embodiment;

[0060] Figure 6 This is a flowchart illustrating the camera management method in another embodiment;

[0061] Figure 7 This is a flowchart illustrating the camera management method in another embodiment;

[0062] Figure 8 This is a structural block diagram of a camera management device in one embodiment;

[0063] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0065] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0066] In Intelligent Traffic Systems (ITS), roadside parking management technology is typically based on images captured by camera equipment, involving cross-technology such as computer vision, IoT device linkage, and smart street light pole integration.

[0067] The deployment of camera equipment can be divided into several methods: ① Low-position cameras (0.5-1.5 meters): This method is easily obstructed by vehicle tires, pedestrians, rain and snow, and the equipment is easily scratched or maliciously damaged, resulting in high maintenance costs; ② High-position cameras (6-8 meters): These require horizontal poles or gantry frames for installation, which is difficult and costly (the cost of a single pole exceeds 10,000 yuan), and the coverage is limited (usually only 4-6 parking spaces can be monitored); ③ Mid-position installation (3.5 meters): Although it can balance cost and coverage (6-8 parking spaces can be monitored), vehicle bodies (such as SUVs and trucks) or surrounding obstacles (such as green belts and road shoulders) can easily obscure license plates, causing the recognition rate to drop to below 70%.

[0068] In related technologies, single-camera, multi-fixed-camera, and pure radar / geomagnetic solutions are commonly used to identify vehicle information for vehicle management. However, single-camera methods cannot simultaneously achieve global monitoring and detailed recognition, with license plate miss rates as high as 30% for non-standard parking (such as angled parking or parking over lines); multi-fixed-camera methods require multiple devices to work together, increasing costs and creating blind spots, making them difficult to adapt to dynamic scenarios; while pure radar or geomagnetic solutions can only detect parking space occupancy status and cannot provide a chain of vehicle characteristic evidence, which can easily lead to management disputes.

[0069] Therefore, when capturing vehicle information, vehicle monitoring is usually based on a dual-camera linkage approach. For example, a fixed surveillance camera (bullet camera) is used to monitor vehicles entering or leaving the vehicle. Upon detecting a vehicle entering or leaving, an integrated PTZ camera is triggered to rotate to a preset position to capture the license plate, thus avoiding license plate obstruction. However, the timing of camera triggering in related technologies is unreasonable, resulting in excessively long working time for the PTZ camera and shortening its lifespan.

[0070] Based on the analysis of the above deployment methods, camera management methods, and other dimensions, this application proposes a dual-camera collaborative working method in a mid-mounted installation scenario. When the first camera detects a target vehicle entering or exiting, it acquires the target vehicle's scene information. If the scene information meets preset scheduling conditions, it acquires the target vehicle's stopping time. If the stopping time exceeds a preset stopping threshold, it instructs the second camera to capture vehicle images of the target vehicle at a preset frequency. The quality of each vehicle image captured by the second camera is evaluated to determine the overall confidence level of each image. Based on the overall confidence level of each vehicle image, the capture frequency of the second camera is adjusted to avoid invalid captures, further shortening the second camera's working time and extending its service life. Simultaneously, it solves the license plate obstruction problem in mid-mounted installation scenarios, optimizing the cost and reliability of traditional camera installation schemes.

[0071] The camera management method provided in this application embodiment can be applied to, for example, Figure 1 The monitoring system shown includes: a controller 102, a first camera device 104 connected to the controller 102, and a second camera device 106. The first camera device 104 can be a fixed-view camera without rotation or zoom capabilities, such as a bullet camera; the second camera device 106 can be a camera with rotation and zoom capabilities, such as an integrated dome camera.

[0072] When the controller detects that the first camera has captured the target vehicle entering or exiting, it acquires the scene information of the target vehicle. If the scene information meets the preset scheduling conditions, it acquires the stopping time of the target vehicle. If the stopping time is greater than the preset stopping threshold, it instructs the second camera to capture vehicle images of the target vehicle at a preset frequency. It then evaluates the quality of each vehicle image captured by the second camera, determines the overall confidence level of each vehicle image, and adjusts the capturing frequency of the second camera based on the overall confidence level of each vehicle image.

[0073] In one exemplary embodiment, such as Figure 2 As shown, a camera management method is provided, applied to a controller in a monitoring system. The monitoring system includes a first camera device and a second camera device connected to the controller. The method includes:

[0074] S201: When the first camera device detects that the target vehicle is entering or leaving, the scene information of the target vehicle is obtained.

[0075] The monitoring system refers to a vehicle monitoring system targeting a specific area, such as a parking lot monitoring system or a roadside monitoring system. The monitoring system is used to manage vehicles within the target area, including monitoring for illegal parking, parking fee collection, and vehicle driving records.

[0076] In practical applications, a surveillance system includes multiple first-level cameras and multiple second-level cameras, with each first-level camera corresponding to and linked with multiple second-level cameras. The first-level cameras are surveillance cameras without rotation or zoom capabilities, such as bullet cameras; each second-level camera is a camera with rotation and zoom capabilities, such as a PTZ camera.

[0077] It should be noted that the shooting function of the second camera device is more complex and the shooting cost is also higher than that of the first camera device. Therefore, in the embodiments of this application, the first camera device is usually in a working state, while each of the second camera devices is usually in a dormant state and will enter the working state only when preset scheduling conditions are met.

[0078] During the operation of the monitoring system, the first camera device is in real-time shooting mode. The controller can monitor whether a target vehicle is entering or leaving in real time based on the images captured by the first camera device, and use the entry or exit of the target vehicle as a trigger node to obtain the current scene information of the target vehicle, and then determine whether the second camera device needs to enter the working state.

[0079] The scene information includes descriptions of the target vehicle's current environment, such as sunny, cloudy, rainy, foggy, bright light, and vehicle obstruction; as well as the device status information of the second camera in the monitoring system, such as flickering, lag, normal operation, blurry lens, and unstable power supply.

[0080] In another application scenario, the first camera device can also send vehicle dynamic information to the controller in a timely manner when it captures a target vehicle entering or exiting. This vehicle dynamic information may include images of the vehicle in motion. Upon receiving the vehicle dynamic information from the first camera device, the controller can obtain scene information of the target vehicle based on the images of the vehicle in motion, and then determine whether the second camera device needs to be activated.

[0081] S202, if the scenario information meets the preset scheduling conditions, then obtain the stopping time of the target vehicle.

[0082] The preset scheduling conditions refer to the standard scene information that allows shooting. Taking the target vehicle's current scene information, including environmental information and the device status information of the second camera, as an example, the scheduling conditions include environmental information of sunny weather and the device status of the second camera being normal.

[0083] When the scene information is multi-dimensional, the scheduling conditions also include multi-dimensional standard scene information. In this embodiment, the scene information meeting the preset scheduling conditions means that all multi-dimensional scene information meets the preset scheduling conditions. Continuing with the example where the scene information includes environmental information and the device status information of the second camera, under the condition that the environmental information is sunny and the device status of the second camera is normal, the stopping time of the target vehicle is obtained.

[0084] When obtaining the parking time of a target vehicle, if the target vehicle is in the entering state, the time interval between the vehicle's entering time and the current time is determined as the parking time of the target vehicle; if the target vehicle is in the exiting state, the time interval between the vehicle's entering time and the vehicle's exiting time is determined as the parking time of the target vehicle.

[0085] In another practical scenario, if the scene information does not meet the preset scheduling conditions, it means that the current scene does not meet the activation requirements of the second camera device, such as inclement weather or unstable power supply to the second camera device. Therefore, to avoid invalid captures by the second camera device, it will not be activated and will not enter working mode, thus eliminating the need to obtain the target vehicle's stopping time.

[0086] S203, if the parking time exceeds the preset parking threshold, instruct the second camera device to capture vehicle images of the target vehicle at a preset frequency.

[0087] The preset parking threshold refers to the upper limit of a vehicle management time period, such as 30 minutes or 60 minutes. If the parking time exceeds the preset parking threshold, it indicates that the target vehicle has been parked for a longer period of time, and more specific vehicle information needs to be obtained for effective management of the target vehicle. In this case, the second camera device is activated and instructed to capture vehicle images of the target vehicle at a preset frequency, such as 30 frames / second or 50 frames / second.

[0088] In one exemplary embodiment, the method further includes: instructing a second camera device to capture a vehicle image of the target vehicle once when the parking time is less than a preset parking threshold.

[0089] If the stopping time is less than the preset stopping threshold, it indicates that the target vehicle's stopping time is short, and no detailed management of the target vehicle is required. In this case, the second camera can still be activated to capture an image of the target vehicle to extract its basic information.

[0090] Furthermore, since the second camera only captures an image once, the license plate area in the vehicle image may be obscured. Therefore, the vehicle image captured by the second camera can be matched with reference vehicle images in the historical database. The license plate information of the reference vehicle image with the highest matching degree can be used as the license plate information of the target vehicle to complete the license plate information of the target vehicle and record it in the database for subsequent tracing.

[0091] S204, perform quality assessment on the vehicle images captured by the second camera device, determine the overall confidence level of each vehicle image, and adjust the capture frequency of the second camera device according to the overall confidence level of each vehicle image, instructing the second camera device to capture vehicle images of the target vehicle at the adjusted capture frequency.

[0092] In actual monitoring scenarios, the second camera device can include multiple cameras, such as multiple PTZ cameras, each of which captures vehicle images of the target vehicle at a preset frequency.

[0093] For each second camera device, during the process of capturing vehicle images at a preset frequency, the quality of each continuously captured vehicle image is evaluated to obtain the overall confidence level of each vehicle image, which serves as a quantitative indicator of the quality of each vehicle image. The higher the overall confidence level of a vehicle image, the higher its quality; conversely, the lower the overall confidence level, the worse its quality.

[0094] In one exemplary embodiment, adjusting the capture frequency of the second camera device based on the overall confidence level of each vehicle image includes:

[0095] If the overall confidence level of all vehicle images is greater than the upper limit of the preset confidence level interval, instruct the second camera to stop capturing images;

[0096] When the overall confidence level of each vehicle image is within the preset confidence level range, the second camera device is instructed to capture images according to the tiered scheduling frequency.

[0097] If the overall confidence level of all vehicle images is less than the lower limit of the preset confidence interval, the second camera device is instructed to stop capturing images.

[0098] Optionally, if the overall confidence level of each vehicle image is greater than the preset confidence level limit, it means that the vehicle images captured by the second camera device are sufficient to extract clear license plate information, and then the second camera device can no longer be instructed to capture images. Correspondingly, the adjusted capture frequency is 0.

[0099] Optionally, if the overall confidence level of all vehicle images is within a preset confidence level range, meaning the vehicle images captured by the second camera are relatively clear and can be processed through post-processing such as stitching and fusion to obtain complete license plate information, then a tiered frequency adjustment can be implemented to reduce the capture frequency of the second camera, instructing it to capture images of the target vehicle at the adjusted capture frequency. It should be noted that the adjusted capture frequency is much lower than the original capture frequency; for example, the capture frequency of the second camera can be adjusted from 50 frames / second to 20 frames / second, 10 frames / second, 5 frames / second, one frame every 15 minutes, and one frame every 30 minutes.

[0100] Optionally, if the overall confidence level of all vehicle images is less than the preset lower confidence level, meaning that the second camera will also have difficulty capturing clear license plate information, then the second camera can stop capturing images. Instead, the second camera captures vehicle images with relatively high overall confidence levels, matches them with reference vehicle images in the historical database, and uses the license plate information of the reference vehicle image with the highest matching degree as the license plate information of the target vehicle to complete the license plate information of the target vehicle. This information is then recorded in the database for subsequent tracing.

[0101] In this embodiment, by comparing the overall confidence level of each vehicle image with a preset confidence interval, when the overall confidence level of each vehicle image is within the preset confidence interval, the capture frequency of the second camera is reduced to optimize the occupation of hardware resources and avoid overuse of the second camera; while when the overall confidence level of each vehicle image is not within the preset confidence interval, the second camera is instructed to stop capturing to save storage space.

[0102] In this embodiment, when the first camera device detects a target vehicle entering or exiting, scene information of the target vehicle is acquired. If the scene information meets preset scheduling conditions, the stopping time of the target vehicle is acquired. If the stopping time exceeds a preset stopping threshold, the second camera device is instructed to capture vehicle images of the target vehicle at a preset frequency. The quality of each vehicle image captured by the second camera device is evaluated to determine the overall confidence level of each vehicle image. Based on the overall confidence level of each vehicle image, the capture frequency of the second camera device is adjusted, and the second camera device is instructed to capture vehicle images of the target vehicle at the adjusted capture frequency. In this method, when a target vehicle is detected entering or exiting, the scene information and vehicle stopping time are further judged. If both meet the scheduling conditions, the second camera device is triggered to start capturing images, reducing the number of times the second camera device is activated, thereby reducing the failure rate of the second camera device and extending its service life. Next, during the capture process of the second camera, the capture frequency of the second camera is adjusted by evaluating the quality of the captured images to avoid invalid captures, further shorten the working time of the second camera, and extend its service life.

[0103] As can be seen from the foregoing embodiments, the controller instructs the second capturing device to initiate snapshot capture based on the premise that the scene information of the target vehicle meets the preset scheduling conditions and the stopping time of the target vehicle meets the preset stopping threshold. Therefore, the accuracy of the scene information and stopping time is crucial to the timing of the second capturing device's snapshot initiation. The foregoing embodiments have already described the steps for obtaining the stopping time in detail; next, the method for obtaining the scene information will be further explained. In an exemplary embodiment, as follows... Figure 3 As shown, the scene information of the target vehicle is obtained, including:

[0104] S301, acquire vehicle images captured by the first camera and historical videos from the second camera.

[0105] The vehicle images captured by the first camera device refer to the images of the target vehicle entering or leaving the vehicle as captured by the first camera device. The historical video of the second camera device refers to the video captured by the second camera device during the most recent historical working period.

[0106] S302, the preset image classification and recognition model is called to analyze the vehicle images collected by the first camera device to obtain environmental information; and the similarity and brightness fluctuation of historical videos are evaluated to determine the device status of the second camera device.

[0107] The image classification and recognition model can be trained using historical vehicle images captured by a first camera device and the environmental labels of each historical vehicle image as a dataset, with an initial classification model, such as the YOLOV8 model, as the training dataset. In practical applications, the input of the image classification and recognition model is the vehicle image, and the output is the environmental category to which the vehicle image belongs, such as black screen, white screen, occlusion, camera occlusion, snow, stripes, color cast, heavy fog, heavy rain, heavy snow, sandstorm, overexposure, underexposure, and occlusion by a cloth.

[0108] For the vehicle image captured by the first camera device, a preset image classification and recognition model is called to analyze the vehicle image captured by the first camera device, identify the environmental area of ​​the input image, and then determine the environmental information of the target vehicle at the current moment.

[0109] For the historical video of the second camera, on the one hand, the similarity of each adjacent frame in the historical video is calculated to obtain the overall similarity of the historical video; on the other hand, the brightness of each video frame in the historical video is calculated, and then the difference between the maximum and minimum brightness values ​​is calculated as the brightness fluctuation amplitude. Then, the brightness fluctuation amplitude corresponding to the historical video is compared with the standard brightness fluctuation amplitude, and the similarity corresponding to the historical video is compared with the standard similarity. Finally, based on the comparison results of the brightness fluctuation amplitude and the similarity comparison results, the device status of the second camera is obtained.

[0110] In an exemplary embodiment, evaluating the similarity and brightness fluctuation amplitude of historical videos to determine the device state of the second camera device includes: acquiring a preset number of adjacent frames from the historical video, and determining the device state of the second camera device as a stuttering state when the structural similarity between each adjacent frame is greater than a similarity threshold; calculating the brightness fluctuation value in each adjacent frame, and determining the device state of the second camera device as a flickering state when the brightness fluctuation value is greater than a preset fluctuation threshold; the brightness fluctuation value includes the maximum brightness fluctuation amplitude value and the maximum brightness fluctuation frequency.

[0111] Structural Similarity Index Measure (SSIM) is a metric that measures the similarity of structure, brightness, and contrast between two images. A higher SSIM indicates a closer similarity between the two images, while a lower SSIM indicates a greater difference. A similarity threshold of 0.95 can be used.

[0112] Taking the acquisition of 6 adjacent frames from historical video as an example, the SSIM between the first and second frames, the SSIM between the second and third frames, the SSIM between the third and fourth frames, the SSIM between the fourth and fifth frames, and the SSIM between the fifth and sixth frames are calculated. Then, the above five SSIMs are compared with the similarity threshold. If all five SSIMs are greater than the similarity threshold, the device status of the second camera is determined to be a stuttering state.

[0113] The brightness fluctuation value and fluctuation frequency of each adjacent frame in the historical video are obtained. The maximum brightness fluctuation value is compared with the preset fluctuation threshold. If the maximum brightness fluctuation value is greater than the preset fluctuation threshold and the fluctuation frequency is greater than 2Hz, the device status of the second camera device is determined to be a flickering state.

[0114] In this embodiment, adjacent video frames of historical videos are evaluated from two dimensions: structural similarity and brightness fluctuation. This provides a comprehensive and reliable reference for evaluating the device status of the second camera device, so as to obtain a more accurate device status of the second camera device.

[0115] S303 uses environmental information and the device status of the second camera as scene information.

[0116] The environmental information and the device status of the second camera are combined to form the scene information of the target vehicle.

[0117] In this embodiment, environmental information of the target vehicle is obtained based on the vehicle image captured by the first camera device, and the device status of the second camera device is obtained based on the historical video of the second camera device. Thus, scene information including both physical environment and device status is provided for triggering the second camera device to work in the future.

[0118] The above embodiments describe the methods for obtaining and judging the scene information and stopping time that need to be determined before instructing the second camera to capture images. Next, the implementation method of determining the comprehensive confidence level based on the captured vehicle image after instructing the second camera to capture images, and then adjusting the capture frequency of the second camera, will be described.

[0119] In one exemplary embodiment, such as Figure 4 As shown, the quality of each vehicle image is assessed to determine the overall confidence level of each vehicle image, including:

[0120] S401, for any vehicle image, determine the quantized value of the vehicle image's size features based on the vehicle image's size feature information, and determine the quantized value of the vehicle image's text features based on the vehicle image's license plate information.

[0121] For any vehicle image, the quantified values ​​of the vehicle image's size features include: vehicle size sharpness, license plate size threshold, and license plate geometric feature value.

[0122] The size sharpness is obtained as follows: the license plate image is resized to the standard reference image size and converted to RGB format; the size similarity score between the converted license plate image and the standard reference image is calculated using the Deep Image Structure and Texture Similarity model (DISTS), and the size sharpness (1 - similarity) is calculated. The higher the size sharpness, the sharper the vehicle image; the lower the size sharpness, the blurrier the vehicle image.

[0123] The license plate size threshold is obtained as follows: Based on the vehicle image, determine the vehicle type and number of rows. Then, based on the mapping relationship between the number of rows and the license plate size, determine the first size threshold (minimum size threshold) and the second size threshold (standard size threshold) for the license plate image. For example, the minimum size threshold (min) for a single row of vehicles is 80×24, and the second size threshold (optimal) is 157×50. If the license plate size is less than the minimum threshold, the license plate size threshold is 0; if the license plate size is greater than or equal to the second size threshold, the license plate size threshold is 1; if the license plate size is between the two size thresholds, calculate the license plate size threshold using linear interpolation. Compare the width of the license plate image with the width of the first size threshold and the width of the second size threshold to determine the width quantization value. Compare the height of the license plate image with the height of the first size threshold and the height of the second size threshold to determine the height quantization value. Finally, determine the minimum value between the width quantization value and the height quantization value as the license plate size threshold.

[0124] The method for obtaining geometric eigenvalues ​​is as follows:

[0125] Calculate the aspect ratio score. The score is calculated based on the difference between the actual aspect ratio of the license plate and the standard aspect ratio. The smaller the difference, the higher the aspect ratio score. The corresponding formula is:

[0126]

[0127] Calculate the rotation angle score. Calculate the horizontal vector using the coordinates of the top left and top right corners of the license plate. Calculate the angle between this horizontal vector and the standard horizontal vector. If the angle is greater than 15°, the rotation angle score is 0. If the angle is less than 15°, the rotation angle score is obtained using the following formula:

[0128]

[0129] The average of the aspect ratio score and the rotation angle score is calculated and determined as the geometric feature value.

[0130] For any vehicle image, the vehicle image is identified to determine the license plate information. Based on each character of the license plate information, the text feature quantization value of the vehicle image is obtained, such as the license plate length feature value (indicating whether the number of characters in the license plate information is consistent with the standard number of characters), the format score (indicating whether the character format in the license plate information matches the standard character format), the single character quantization value (indicating the accuracy score of a specified character in the license plate information), and the obfuscated character quantization value (indicating the accuracy score of each character in the license plate information).

[0131] S402, weighted fusion of size feature quantization values ​​and text feature quantization values, to obtain the overall confidence level of the vehicle image.

[0132] Taking the size feature quantization values, including vehicle size clarity, license plate size threshold, and license plate geometric feature values, and the text feature quantization values, including license plate length feature values, format score, single character quantization value, and obfuscated character quantization value, as an example, weights are pre-assigned to each quantization value, and the size feature quantization values ​​and text feature quantization values ​​are weighted and fused according to the weight values ​​of each type of quantization value to obtain the comprehensive confidence of the vehicle image.

[0133] In practical applications, license plate recognition can be performed on vehicle images corresponding to the highest confidence level to determine the license plate number of the target vehicle.

[0134] In this embodiment, for any vehicle image, the size feature quantization value of the vehicle image is determined based on the size feature information of the vehicle image, and the text feature quantization value of the vehicle image is determined based on the license plate information of the vehicle image. The size feature quantization value and the text feature quantization value are weighted and fused to obtain the comprehensive confidence level of the vehicle image, accurately quantifying the quality of the vehicle image and providing a reliable basis for subsequent evaluation of the shooting effect of the second camera device.

[0135] In an exemplary embodiment, the license plate information includes a license plate string; one possible implementation of the aforementioned S401 step of "determining the text feature quantization value of the vehicle image based on the license plate information of the vehicle image" is described below, such as... Figure 5 As shown, it includes:

[0136] S501, determine the character at the preset position from the license plate string.

[0137] It should be noted that license plate strings are typically composed of elements from a pre-defined Chinese character library, letter library, and number library. For example, the first character in a license plate string might be a Chinese character, the second character a letter, and the remaining characters might be numbers or letters.

[0138] In this embodiment of the application, the character in the license plate string at a preset position can be the second character determined according to the left-to-right order of the license plate string.

[0139] S502, match the character in the license plate string at the preset position with each character in the first character library to determine the quantization value of a single character.

[0140] The first character library includes various characters that might appear in real-world application scenarios, with the characters at preset positions. Continuing with the example of the second character, whose preset position is determined by the left-to-right order of the license plate string, the second character is a letter. Correspondingly, the first character library is the alphabet library, including 24 of the 26 letters except for I / O (AH, JN, PZ).

[0141] Matching the character at a preset position in the license plate string with each character in the first character library means determining whether the character at the preset position exists in the first character library. If a reference character matching the character exists in the first character library, the quantization value of the single character is determined to be 1, and the image quality is high; if no reference character matching the character exists in the first character library, the quantization value of the single character is determined to be 0, and the image quality is low.

[0142] S503, the quantization value of a single character is determined as the quantization value of the text feature.

[0143] In this embodiment, the character at a preset position in the license plate string is matched with each character in the first character library to determine the quantization value of a single character, and then the quantization value of the text feature is determined. In this way, by focusing on key characters, the efficient quantization of the license plate text feature is achieved.

[0144] In an exemplary embodiment, the license plate information includes a license plate string; one possible implementation of the aforementioned S401 step of "determining the text feature quantization value of the vehicle image based on the license plate information of the vehicle image" is described below, such as... Figure 6 As shown, it includes:

[0145] S601, match each character in the license plate string with each character in the second character library to determine the credibility of each character in the license plate string.

[0146] It is important to reiterate that the license plate string includes Chinese characters, letters, and numbers. For example, the first character in the license plate string might be a Chinese character, the second a letter, and the remaining characters might be numbers or letters. Therefore, when the specific type of each character in the license plate string cannot be determined, matching each character requires matching it with all possible characters. In this embodiment, all possible characters in the license plate string are summarized to obtain a second character library, which includes an letter library, a number library, a Chinese character library, and a background.

[0147] It should be noted that, in this embodiment of the application, the relationship between the second character library and the first character library is as follows: the first character library is a character library set for characters at preset positions, and the second character library is a character library set for all characters in the license plate string. The second character library includes the first character library.

[0148] In an exemplary embodiment, each character in the license plate string is matched with characters in a second character library to determine the credibility of each character in the license plate string, including:

[0149] For each character in the license plate string, calculate the similarity between the character and each character in the second character library, and map the maximum similarity and the second maximum similarity to obtain the credibility of the character.

[0150] For a single character, the higher the confidence value, the more unique and unconfused the recognition result is; the lower the confidence value, the worse the recognition effect is and the greater the confusion.

[0151] Taking the second character library, which includes 78 characters (44 Chinese characters + 24 letters + 10 numbers) and 1 background, as an example, for each character in the license plate string, the similarity between the character and the 78 characters is calculated, including the 78 similarities. Then, the 78 similarities are sorted, and the maximum similarity and the second maximum similarity are mapped to the confidence level through a modified mapping function (such as the Sigmoid function).

[0152] In this embodiment, the similarity between each character of the license plate and each character in the second character library is calculated, and the maximum and second largest similarity are mapped to obtain the single character confidence. This accurately captures the distinguishability of a single character from similar characters and provides an effective evaluation basis for the recognition effect of a single character.

[0153] S602, by comparing the credibility of each character in the license plate string, the minimum credibility is determined as the quantization value of the obfuscated character.

[0154] After obtaining the credibility of each character in the license plate string, the credibility is sorted, and the minimum credibility, which is the quantification index of the worst character recognition effect, is determined as the quantification value of the obfuscated character.

[0155] S603, the quantization value of the obfuscated character is determined as the quantization value of the text feature.

[0156] In this embodiment, each character in the license plate string is matched with each character in the second character library to determine the credibility of each character in the license plate string. The minimum credibility is determined as the quantification value of the obfuscated character. By accurately capturing the weakest low-credibility characters in the license plate, the weak links in license plate character recognition are highlighted, and the recognition interference caused by obfuscated characters is effectively quantified. This provides targeted feature support for improving the accuracy of license plate recognition in the future by determining the text feature quantification value.

[0157] In an exemplary embodiment, the license plate information includes the license plate length and license plate format; one possible implementation method of "determining the text feature quantization value of the vehicle image based on the license plate information of the vehicle image" in the aforementioned S401 is described, such as... Figure 7 As shown, it includes:

[0158] S701 performs color recognition on the license plate information of the vehicle image to determine the vehicle type of the target vehicle.

[0159] The vehicle type of the target vehicle can be identified by recognizing the license plate color information of the vehicle image, including but not limited to blue-plate motor vehicles, yellow-plate motor vehicles, black-plate motor vehicles, white-plate motor vehicles, green-plate motor vehicles, and electric vehicles.

[0160] S702, based on the vehicle type of the target vehicle, determine the standard length and standard format of the license plate of the target vehicle.

[0161] The standard length of a license plate refers to the string length. The standard length of a license plate for blue, yellow, black, and white vehicles is 7 characters. The standard length of a license plate for green vehicles is 8 characters. The standard length of a license plate for electric vehicles is 6 characters.

[0162] The standard format for license plates for blue, yellow, black, and white vehicles is: one Chinese character, one letter, and a five-character string sequence (number + letter). The standard format for green license plates is: one Chinese character, one letter, and a six-character string sequence (number + letter). The standard format for electric vehicle license plates is: one Chinese character, one letter, and a four-character string sequence (number + letter). It should be noted that the numbers in the string sequence refer to the digits 0-9, and the letters refer to all letters of the alphabet except I and O.

[0163] S703, compare the standard license plate length with the license plate length to determine the license plate length feature value of the vehicle image; and match the standard license plate format with the license plate format to determine the license plate format feature value of the vehicle image.

[0164] If the standard license plate length matches the vehicle license plate length, the license plate length feature value is set to 1; if they do not match, the license plate length feature value is set to 0. Regular expressions are used to verify whether the standard license plate format matches the vehicle license plate format's text format. If they match, the license plate format feature value is set to 1; otherwise, it is set to 0.

[0165] S704, based on the license plate length feature value and the license plate format feature value, determines the text feature quantization value of the vehicle image.

[0166] Both the license plate length feature value and the license plate format feature value are used as the text feature quantization values ​​of the vehicle image.

[0167] In this embodiment, the vehicle type is identified by license plate color recognition, and then the corresponding standard license plate length and format are matched. The license plate length consistency and format matching degree are then quantized by binary 0 and 1 respectively. Finally, the two types of feature values ​​are fused to obtain the text feature quantization value. This achieves accurate and efficient quantization of license plate text features based on vehicle type adaptation rules, which not only ensures the pertinence and standardization of feature quantization, but also provides simple and reliable feature support for subsequent license plate verification, recognition result verification and other scenarios.

[0168] In one exemplary embodiment, a specific camera management method is provided, applied to a controller in a surveillance system, the surveillance system including bullet cameras and PTZ cameras connected to the controller; the method includes:

[0169] (1) When the camera detects that the target vehicle is entering or leaving the camera, acquire the vehicle image captured by the camera and the historical video of the PTZ camera.

[0170] (2) Call the preset image classification and recognition model to analyze the vehicle images collected by the first camera device to obtain environmental information; and evaluate the similarity and brightness fluctuation of historical videos to determine the device status of the PTZ camera.

[0171] (3) Under the condition that the environmental information is sunny or bright, and the PTZ camera is in normal condition, obtain the parking time of the target vehicle.

[0172] (4) If the stopping time is less than the preset stopping threshold, instruct the PTZ camera to capture a vehicle image of the target vehicle once.

[0173] (5) When the parking time exceeds the preset parking threshold, instruct the PTZ camera to capture vehicle images of the target vehicle at a preset frequency.

[0174] It should be noted that "PTZ camera" refers to multiple backup PPT cameras located 1-2 poles adjacent to the main camera based on the pole position topology relationship on the electronic map. The PPT cameras are instructed to take the shot to avoid mechanical damage caused by forced linkage of a single-pole device.

[0175] Furthermore, based on the exiting vehicle, the system is linked to the adjacent left and right poles to capture the license plates of the vehicle before and after it. That is, the adjacent pole on the right captures the license plate of the vehicle in the parking space before it exits, and the adjacent pole on the left captures the license plate of the vehicle in the parking space after it exits. At the same time, it determines whether the parking spaces in front and behind are already parked. If they are not parked, the rotating PTZ camera is not triggered to capture the license plate.

[0176] (6) Evaluate the quality of each vehicle image captured by the PTZ camera and determine the overall confidence level of each vehicle image.

[0177] For any vehicle image, seven quality index quantization values ​​are determined from seven dimensions: vehicle size clarity, license plate size threshold, license plate geometric feature value, license plate length feature value, format score, single character quantization value, and obfuscated character quantization value. Then, according to the preset weighting, the quantization values ​​of each quality index are weighted to obtain the overall confidence level of the vehicle image.

[0178] (7) When the overall confidence level of each vehicle image is greater than the upper limit of the preset confidence interval, instruct the PTZ camera to stop capturing images.

[0179] (8) When the overall confidence level of each vehicle image is within the preset confidence level range, the PTZ camera is instructed to capture images according to the tiered scheduling frequency.

[0180] (9) If the overall confidence level of each vehicle image is less than the lower limit of the preset confidence interval, instruct the PTZ camera to stop capturing images.

[0181] In practical applications, the reliability of a single character and the overall license plate quality can be calculated for each license plate recognition iteration, and then compared with the previous reliability of a single character and the overall license plate quality. If the current reliability of a single character is greater than the previous single character reliability, the current reliability of the single character and the character recognition result are updated, and the current overall license plate reliability is updated according to the formula for overall license plate quality reliability.

[0182] The management method provided in this application has the following advantages:

[0183] First, the lifespan and reliability of the PTZ camera have been significantly improved.

[0184] Mechanical wear and tear is significantly reduced: Through scene classification and three-level prediction, the average number of daily linkages of PTZ cameras has been reduced from more than 200 times to 50-80 times (within the reasonable design threshold), the rotation cycle life has been extended from 1-1.5 years to 4-5 years (close to the upper limit of the design life), and the replacement cost of a single device has been reduced by more than 70%.

[0185] Reduced failure rate: Reduced wear on the PTZ caused by high-frequency rotation, reducing the failure rate of PTZ cameras such as jamming and angle deviation from 40% to below 10%, and increasing the effective working time of the PTZ camera to over 98%.

[0186] Second, optimize system resource utilization efficiency.

[0187] Invalid linkage filtering: Environment classification and event grading can directly reduce redundant linkages by 60%-70%, reduce edge node computing power consumption by 40%, and reduce the average daily bandwidth consumption per pole from 15GB to 8GB.

[0188] Cross-pole collaboration improves coverage: Through dynamic resource scheduling, the license plate recognition coverage in the traditional single-pole blind spot is increased from 60% to 95%, avoiding management interruptions caused by single-pole failure.

[0189] Third, improved identification efficiency and management experience.

[0190] Enhanced accuracy and stability: The linkage is triggered only in effective scenarios, improving the license plate recognition accuracy from a fluctuating 60%-90% to a stable 95% or more, and reducing the reliance on the cross-time domain supplementation mechanism from 25% to 5%.

[0191] Structural reduction in operation and maintenance costs: Reduced frequency of PTZ camera replacements and a decrease in manual verification rate (from 15% to 2%) have reduced overall operation and maintenance costs by 50%-60%. Meanwhile, due to the precision of the linkage strategy, the success rate of seamless payment for users has increased to 99%, and the complaint rate has decreased by 95%.

[0192] In summary, by adopting the management logic of "classification and prediction to reduce invalid actions, multi-level scheduling to balance performance and losses, and feedback optimization to continuously adapt to scenarios" in the embodiments of this application, the core pain points of traditional PTZ camera linkage such as short lifespan, low efficiency, and high cost are solved, providing a highly reliable and low-cost linkage solution for mid-position dual-camera systems.

[0193] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0194] Based on the same inventive concept, this application also provides a camera management device for implementing the camera management method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more camera management device embodiments provided below can be found in the limitations of the camera management method described above, and will not be repeated here.

[0195] This application also provides a monitoring system, which includes: a controller, a first camera device and a second camera device connected to the controller; the controller is used to execute the steps of the method in any of the above-described camera management methods.

[0196] In one exemplary embodiment, such as Figure 8 As shown, a camera management device is provided, including: a scene information acquisition module 801, a docking time acquisition module 802, a docking time comparison module 803, and a capture frequency adjustment module 804, wherein:

[0197] The scene information acquisition module 801 is used to acquire scene information of the target vehicle when the first camera device captures the target vehicle entering or leaving the vehicle.

[0198] The stopping time acquisition module 802 is used to acquire the stopping time of the target vehicle if the scenario information meets the preset scheduling conditions.

[0199] The stopping time comparison module 803 is used to instruct the second camera device to capture vehicle images of the target vehicle at a preset frequency when the stopping time is greater than a preset stopping threshold.

[0200] The capture frequency adjustment module 804 is used to evaluate the quality of each vehicle image captured by the second camera device, determine the comprehensive confidence level of each vehicle image, and adjust the capture frequency of the second camera device according to the comprehensive confidence level of each vehicle image, instructing the second camera device to capture vehicle images of the target vehicle at the adjusted capture frequency.

[0201] In an exemplary embodiment, the scene information acquisition module 801 includes: an image and video acquisition unit, an image and video analysis unit, and a scene information aggregation unit, wherein:

[0202] The image and video acquisition unit is used to acquire vehicle images captured by the first camera device and historical videos from the second camera device;

[0203] The image and video analysis unit is used to call a preset image classification and recognition model to analyze the vehicle images captured by the first camera device to obtain environmental information; and to evaluate the similarity and brightness fluctuation of historical videos to determine the device status of the second camera device.

[0204] The scene information aggregation unit is used to collect environmental information and the device status of the second camera as scene information.

[0205] In an exemplary embodiment, the image and video analysis unit is further configured to obtain a preset number of adjacent frames from historical videos, and determine that the device state of the second camera device is a stuttering state when the structural similarity between each adjacent frame is greater than a similarity threshold; calculate the brightness fluctuation value in each adjacent frame, and determine that the device state of the second camera device is a flickering state when the brightness fluctuation value is greater than a preset fluctuation threshold; the brightness fluctuation value includes the maximum brightness fluctuation amplitude value and the maximum brightness fluctuation frequency.

[0206] In an exemplary embodiment, the capture frequency adjustment module 804 further includes: a feature quantization value determination unit and a confidence level determination unit, wherein:

[0207] The feature quantization value determination unit is used to determine the size feature quantization value of any vehicle image based on the size feature information of the vehicle image, and to determine the text feature quantization value of the vehicle image based on the license plate information of the vehicle image.

[0208] The confidence determination unit is used to perform weighted fusion of the quantized values ​​of size features and text features to obtain the comprehensive confidence of the vehicle image.

[0209] In an exemplary embodiment, the license plate information includes a license plate string; the feature quantization value determination unit is further configured to determine a character at a preset position from the license plate string; match the character at the preset position in the license plate string with each character in the first character library to determine a single character quantization value; and determine the single character quantization value as a text feature quantization value.

[0210] In an exemplary embodiment, the license plate information includes a license plate string; the feature quantization value determination unit is further configured to match each character in the license plate string with each character in the second character library to determine the credibility of each character in the license plate string; by comparing the credibility of each character in the license plate string, the minimum credibility is determined as the obfuscated character quantization value; and the obfuscated character quantization value is determined as the text feature quantization value.

[0211] In an exemplary embodiment, the feature quantization value determination unit is further configured to calculate the similarity between each character in the license plate string and each character in the second character library; and map the maximum similarity and the second maximum similarity to obtain the credibility of the character.

[0212] In an exemplary embodiment, the license plate information includes the license plate length and license plate format; the feature quantization value determination unit is further configured to perform color recognition on the license plate information of the vehicle image to determine the vehicle type of the target vehicle; based on the vehicle type of the target vehicle, determine the standard length and standard format of the license plate of the target vehicle; compare the standard length of the license plate with the license plate length to determine the license plate length feature value of the vehicle image; and match the standard format of the license plate with the license plate format to determine the license plate format feature value of the vehicle image; and based on the license plate length feature value and the license plate format feature value, determine the text feature quantization value of the vehicle image.

[0213] In an exemplary embodiment, the capture frequency adjustment module 804 includes: a first adjustment unit, a second adjustment unit, and a third adjustment unit, wherein:

[0214] The first adjustment unit is used to instruct the second camera device to stop capturing images when the overall confidence level of each vehicle image is greater than the upper limit of the preset confidence level interval.

[0215] The second adjustment unit is used to instruct the second camera to capture images according to the tiered scheduling frequency when the overall confidence level of each vehicle image is within the preset confidence level range.

[0216] The third adjustment unit is used to instruct the second camera to stop capturing images when the overall confidence level of each vehicle image is less than the lower limit of the preset confidence level interval.

[0217] In one exemplary embodiment, the camera management device includes a frequency fine-tuning module, configured to instruct a second camera device to capture a vehicle image of the target vehicle once when the parking time is less than a preset parking threshold.

[0218] Each module in the aforementioned camera management device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0219] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores camera management data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a camera management method.

[0220] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0221] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0222] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0223] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0224] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0225] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0226] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0227] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A camera management method, characterized in that, A controller used in a monitoring system, the monitoring system including a first camera device and a second camera device connected to the controller; the method includes: If the first camera device detects that a target vehicle is entering or exiting, the scene information of the target vehicle is obtained. If the scenario information meets the preset scheduling conditions, then the stopping time of the target vehicle is obtained; If the stopping time exceeds a preset stopping threshold, the second camera device is instructed to capture vehicle images of the target vehicle at a preset frequency; The quality of each vehicle image captured by the second camera device is evaluated to determine the overall confidence level of each vehicle image. Based on the overall confidence level of each vehicle image, the capture frequency of the second camera device is adjusted, and the second camera device is instructed to capture vehicle images of the target vehicle at the adjusted capture frequency.

2. The method according to claim 1, characterized in that, The acquisition of the scene information of the target vehicle includes: Acquire vehicle images captured by the first camera device and historical videos from the second camera device; The system calls a preset image classification and recognition model to analyze the vehicle images captured by the first camera device to obtain environmental information; and evaluates the similarity and brightness fluctuation of the historical video to determine the device status of the second camera device. The environmental information and the device status of the second camera device are used as the scene information.

3. The method according to claim 2, characterized in that, The step of evaluating the similarity and brightness fluctuation range of the historical videos to determine the device status of the second camera includes: A preset number of adjacent frames are obtained from the historical video, and if the structural similarity between each of the adjacent frames is greater than the similarity threshold, the device status of the second camera device is determined to be a stuttering state. Calculate the brightness fluctuation value in each of the adjacent frames, and determine the device state of the second camera device as a flickering state if the brightness fluctuation value is greater than a preset fluctuation threshold; the brightness fluctuation value includes the maximum brightness fluctuation amplitude value and the maximum brightness fluctuation frequency.

4. The method according to any one of claims 1-3, characterized in that, The process of evaluating the quality of each vehicle image and determining the overall confidence level of each vehicle image includes: For any vehicle image, the size feature quantization value of the vehicle image is determined based on the size feature information of the vehicle image, and the text feature quantization value of the vehicle image is determined based on the license plate information of the vehicle image. The quantized values ​​of the size features and the quantized values ​​of the text features are weighted and fused to obtain the overall confidence level of the vehicle image.

5. The method according to claim 4, characterized in that, The license plate information includes a license plate string; determining the text feature quantization value of the vehicle image based on the license plate information includes: Determine the character at the preset position from the license plate string; The character at a preset position in the license plate string is matched with each character in the first character library to determine the quantization value of a single character; The single-character quantization value is determined as the text feature quantization value.

6. The method according to claim 4, characterized in that, The license plate information includes a license plate string; determining the text feature quantization value of the vehicle image based on the license plate information includes: Each character in the license plate string is matched with each character in the second character library to determine the credibility of each character in the license plate string; By comparing the credibility of each character in the license plate string, the minimum credibility is determined as the quantization value of the obfuscated character; The quantization value of the obfuscated character is determined as the quantization value of the text feature.

7. The method according to claim 6, characterized in that, The step of matching each character in the license plate string with characters in the second character library to determine the credibility of each character in the license plate string includes: For each character in the license plate string, calculate the similarity between the character and each character in the second character library; The credibility of the character is obtained by mapping the maximum similarity and the second maximum similarity.

8. The method according to claim 4, characterized in that, The license plate information includes the license plate length and license plate format; determining the text feature quantization value of the vehicle image based on the license plate information of the vehicle image includes: Color recognition is performed on the license plate information of the vehicle image to determine the vehicle type of the target vehicle; Based on the vehicle type of the target vehicle, determine the standard length and standard format of the license plate of the target vehicle; The standard length of the license plate is compared with the license plate length to determine the license plate length feature value of the vehicle image; and the standard format of the license plate is matched with the license plate format to determine the license plate format feature value of the vehicle image. Based on the license plate length feature value and the license plate format feature value, the text feature quantization value of the vehicle image is determined.

9. The method according to any one of claims 1-3, characterized in that, The step of adjusting the capture frequency of the second camera device based on the comprehensive confidence level of each vehicle image includes: If the overall confidence level of all the vehicle images is greater than the upper limit of the preset confidence interval, the second camera device is instructed to stop capturing images. When the overall confidence level of each vehicle image is within the preset confidence level range, the second camera device is instructed to capture images according to the tiered scheduling frequency. If the overall confidence level of all the vehicle images is less than the lower limit of the preset confidence interval, the second camera device is instructed to stop capturing images.

10. The method according to any one of claims 1-3, characterized in that, The method further includes: If the stopping time is less than the preset stopping threshold, the second camera device is instructed to capture a vehicle image of the target vehicle once.

Images