Image quality adaptive adjustment method and system in high-definition video surveillance system
By extracting basic feature information, detecting noise categories, identifying and tracking targets in high-definition video surveillance system, adaptive adjustment of image quality is achieved, and the problem that traditional methods cannot adapt to complex environments and inefficiency is solved, and image quality and monitoring are improved Effect.
Patent Information
- Application Number
- CN202411028663.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-07-30
AI Technical Summary
The image quality adjustment method of traditional high-definition video surveillance systems relies on fixed parameter settings and manual intervention, and cannot adapt to complex and changeable environmental conditions, is inefficient, and is difficult to make accurate adjustments in real time.
By obtaining the original image in the high-definition video surveillance system, extracting basic feature information such as brightness, contrast, color saturation and clarity, analyzing image quality, detecting noise categories, enhancing image quality, identifying detection targets, and building an initial image tracing frame. Perform scheduling adjustments, track target trajectory, analyze behavior patterns, and dynamic adjustments to image quality.
It realizes adaptive adjustment of image quality of high-definition video surveillance system in various scenarios, improves the clarity and accuracy of the surveillance image, can quickly solve image quality problems, clarify target categories, continuously track targets, intuitively display motion conditions, and evaluate Image quality adjustment level.
Smart Images

Figure CN118552458B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image quality adjustment, and in particular to a method and system for adaptively adjusting image quality in a high-definition video monitoring system. Background Art
[0002] In today's security monitoring field, high-definition video surveillance systems are increasingly widely used, playing an important role in ensuring public safety and enterprise management. In particular, in terms of image quality optimization, image quality adaptive adjustment methods have attracted much attention because they can dynamically optimize display effects according to environmental and scene changes.
[0003] At present, traditional methods for adjusting the image quality of high-definition video surveillance systems mainly rely on fixed parameter settings and manual intervention. Although these methods can guarantee a certain image quality in certain specific scenarios, they have obvious limitations. Among them, fixed parameter settings cannot adapt to complex and changing environmental conditions, such as different light intensities, dynamic scene changes, etc.; manual intervention is not only inefficient, but also difficult to accurately adjust the image quality in real time. Therefore, an image quality adaptive adjustment method based on advanced technology is needed to improve the image quality of high-definition video surveillance systems in various scenarios. Summary of the invention
[0004] The present invention provides a method and system for adaptively adjusting image quality in a high-definition video monitoring system, the main purpose of which is to improve the monitoring image quality of the high-definition video monitoring system.
[0005] To achieve the above object, the present invention provides a method for adaptively adjusting image quality in a high-definition video surveillance system, comprising:
[0006] Acquire an original image in a high-definition video surveillance system, extract basic feature information of the original image, and analyze the image quality of the original image based on the basic feature information, wherein the basic feature information includes brightness, contrast, color saturation and clarity;
[0007] When the image quality does not meet the preset quality, detecting the noise category of the original image, based on the noise category, performing image quality enhancement on the original image to obtain an enhanced image, and when the enhanced image quality corresponding to the enhanced image meets the preset quality, obtaining a first adjusted image;
[0008] Identify a detection target of the first adjustment image, construct an initial image frame of the first adjustment image based on the detection target, identify feature points of the detection target, and perform frame adjustment on the initial image frame based on the feature points to obtain a target frame;
[0009] Tracking the detection target using the target frame to obtain a target trajectory, analyzing a behavior pattern of the detection target based on the target trajectory, identifying the clarity of the detection target when the behavior pattern is abnormal, and performing a second adjustment on the first adjustment image to obtain a second adjustment image when the clarity does not meet a preset requirement;
[0010] An image adjustment level of the second adjustment image is analyzed, and when the image adjustment level meets a preset level, the second adjustment image is used as a target adjustment image.
[0011] Optionally, analyzing the image quality of the original image based on the basic feature information includes:
[0012] Constructing weights corresponding to the basic feature information to obtain brightness evaluation weights, contrast weights, color saturation weights, and clarity weights;
[0013] Based on the brightness evaluation weight, the contrast weight, the color saturation weight, and the clarity weight, the image quality of the original image is calculated in combination with the following formula:
[0014]
[0015] Wherein, Q represents the image quality, L represents the brightness evaluation value corresponding to the basic feature information, C represents the contrast evaluation value corresponding to the basic feature information, S represents the color saturation evaluation value corresponding to the basic feature information, D represents the clarity evaluation value corresponding to the basic feature information, represents the brightness evaluation weight, represents the contrast weight, Indicates the color saturation weight, Indicates the clarity weight.
[0016] Optionally, the detecting the noise category of the original image includes:
[0017] Performing an original two-dimensional discrete Fourier transform on the original image to obtain frequency domain data about the original image;
[0018] Querying the amplitude spectrum and phase spectrum in the frequency domain data;
[0019] Analyzing frequency domain features of the original image based on the amplitude spectrum and the phase spectrum;
[0020] The noise category of the original image is determined based on the frequency domain features.
[0021] Optionally, the performing image quality enhancement on the original image based on the noise category to obtain an enhanced image includes:
[0022] Adaptively denoising the original image based on the noise category to obtain a denoised image;
[0023] Performing sharpening processing on the denoised image to obtain a sharpened image;
[0024] Calculating a brightness distribution histogram of the sharpened image;
[0025] Performing brightness adjustment on the sharpened image based on the calculation result of the brightness distribution histogram to obtain a brightness adjusted image;
[0026] Color correction is performed on the brightness-adjusted image to obtain an enhanced image.
[0027] Optionally, constructing an initial image frame of the first adjusted image based on the detected target includes:
[0028] Calculating the target size of the detection target;
[0029] Constructing a frame size of the initial image based on the target size;
[0030] Identifying position distribution information of the detection target in the first adjusted image;
[0031] Based on the position distribution information, construct a frame distribution strategy for the initial image;
[0032] An initial image frame of the first adjusted image is constructed based on the frame size and the frame distribution strategy.
[0033] Optionally, identifying feature points of the detection target includes:
[0034] Constructing a scale space of the detection target;
[0035] Marking key points of the detection target in the scale space to obtain marked key points;
[0036] Based on the marked key points, after the directions of the marked key points are assigned, the marked key points are characterized by using the following formula to obtain the feature points:
[0037] ;
[0038] in, Represents feature points, Indicates the image gradient value of the detected target corresponding to the first adjusted image, Indicates the direction of the marked key point, represents the feature image, , , , Indicates the origin position of the feature image in the scale space The nearest neighboring points.
[0039] Optionally, the assigning directions to the marked key points includes:
[0040] Calculating the domain gradient of the key point;
[0041] Constructing a direction histogram of the marked key points according to the gradient direction corresponding to the domain gradient;
[0042] The peak value in the direction histogram is used as the direction of the marked key point.
[0043] Optionally, adjusting the initial image frame based on the feature points to obtain a target frame includes:
[0044] Based on the feature points, construct a real bounding box of the first adjusted image;
[0045] Calculating the intersection-over-union ratio of the real bounding box and the initial image frame;
[0046] The initial image frame is adjusted based on the intersection-and-union comparison to obtain a target frame.
[0047] Optionally, analyzing the behavior pattern of the detection target based on the target trajectory includes:
[0048] Extracting behavioral features of the target trajectory;
[0049] Based on the behavior characteristics, the target trajectory is visualized to obtain a visualized trajectory;
[0050] Performing behavioral semantic analysis on the visualization trajectory to obtain a behavioral pattern of the detection target.
[0051] In order to solve the above problems, the present invention also provides an image quality adaptive adjustment system in a high-definition video monitoring system, the system comprising:
[0052] An image quality analysis module is used to obtain an original image in a high-definition video surveillance system, extract basic feature information of the original image, and analyze the image quality of the original image based on the basic feature information, wherein the basic feature information includes brightness, contrast, color saturation and clarity;
[0053] a first image adjustment module, configured to detect a noise category of the original image when the image quality does not meet a preset quality, enhance the image quality of the original image based on the noise category to obtain an enhanced image, and obtain a first adjusted image when the enhanced image quality corresponding to the enhanced image meets the preset quality;
[0054] a frame adjustment module, configured to identify a detection target of the first adjustment image, construct an initial image frame of the first adjustment image based on the detection target, identify feature points of the detection target, and perform frame adjustment on the initial image frame based on the feature points to obtain a target frame;
[0055] an image second adjustment module, configured to track the detection target using the target trace frame to obtain a target trajectory, analyze a behavior pattern of the detection target based on the target trajectory, identify the clarity of the detection target when the behavior pattern is abnormal, and perform a second adjustment on the first adjustment image to obtain a second adjustment image when the clarity does not meet a preset requirement;
[0056] The target image acquisition module is used to analyze the image adjustment level of the second adjustment image, and when the image adjustment level meets a preset level, use the second adjustment image as a target adjustment image.
[0057] The present invention can analyze the current monitoring effect of the high-definition video surveillance system by acquiring the original image in the high-definition video surveillance system; the present invention can understand the specific reasons affecting the image quality by detecting the noise category of the original image when the image quality does not meet the preset quality, and then quickly solve the problem and improve the image quality; the present invention can clarify the specific category of each target in the image by identifying the detection target of the first adjustment image, such as distinguishing whether it is a person, an animal or an object, and then continuously track the detected target in a continuous image sequence to understand its motion trajectory and behavior; the present invention can visualize the detection target by tracking the detection target using the target frame to obtain the target trajectory, and then more intuitively present the movement of the target, which is convenient for users to observe and analyze; finally, the present invention can evaluate whether the quality of the processed image is improved or reduced compared with the original image by analyzing the image adjustment level of the second adjustment image, and analyze the quality loss of the image during compression transmission. Therefore, the present invention can improve the monitoring image quality of the high-definition video surveillance system. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] Figure 1 A schematic diagram of a flow chart of a method for adaptively adjusting image quality in a high-definition video surveillance system provided by an embodiment of the present invention;
[0059] Figure 2 A functional module diagram of an image quality adaptive adjustment system in a high-definition video surveillance system provided by an embodiment of the present invention;
[0060] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0061] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.
[0062] The embodiment of the present application provides a method for adaptively adjusting image quality in a high-definition video surveillance system. The execution subject of the method for adaptively adjusting image quality in a high-definition video surveillance system includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided in the embodiment of the present application. In other words, the method for adaptively adjusting image quality in a high-definition video surveillance system can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms.
[0063] Reference Figure 1 FIG. 1 is a flow chart of a method for adaptively adjusting image quality in a high-definition video surveillance system provided by an embodiment of the present invention. In this embodiment, the method for adaptively adjusting image quality in a high-definition video surveillance system includes:
[0064] S1. Acquire an original image in a high-definition video surveillance system, extract basic feature information of the original image, and analyze the image quality of the original image based on the basic feature information, wherein the basic feature information includes brightness, contrast, color saturation and clarity.
[0065] The embodiment of the present invention can analyze the current monitoring effect of the high-definition video surveillance system by acquiring the original image in the high-definition video surveillance system. The high-definition video surveillance system refers to a security system that uses high-definition cameras and video processing technology to monitor and record specific areas in real time. The high-definition video surveillance system usually includes components such as cameras, video recorders, video transmission equipment, and monitoring software, and can realize functions such as remote monitoring, alarm linkage, image storage and retrieval.
[0066] Furthermore, the embodiment of the present invention can specifically understand whether the quality of the image currently available to the high-definition video surveillance system does not meet the requirements and what problems exist by extracting the basic feature information of the original image. The basic feature information includes brightness, contrast, color saturation and clarity.
[0067] The brightness can be calculated by the average brightness method, the contrast can be calculated by the standard deviation method, the color saturation can be calculated by the RGB channel method, and the clarity can be identified by converting the image into the frequency domain through Fourier transform, and the energy of the high-frequency part is related to the clarity.
[0068] The embodiment of the present invention can understand whether the camera picture obtained under the original setting parameters of the high-definition video surveillance system is clear enough by analyzing the image quality of the original image based on the basic feature information. If specific target analysis is required, it can meet the needs, such as analyzing whether a person in the image is clear enough.
[0069] As an embodiment of the present invention, analyzing the image quality of the original image based on the basic feature information includes: constructing weights corresponding to the basic feature information to obtain brightness evaluation weights, contrast weights, color saturation weights and clarity weights, and calculating the image quality of the original image based on the brightness evaluation weights, the contrast weights, the color saturation weights and the clarity weights in combination with the following formula:
[0070] ;
[0071] Wherein, Q represents the image quality, L represents the brightness evaluation value corresponding to the basic feature information, C represents the contrast evaluation value corresponding to the basic feature information, S represents the color saturation evaluation value corresponding to the basic feature information, D represents the clarity evaluation value corresponding to the basic feature information, represents the brightness evaluation weight, represents the contrast weight, Indicates the color saturation weight, Indicates the clarity weight.
[0072] Among them, the weights corresponding to the basic feature information are constructed according to the needs and experience in practical applications. For example, in some cases, brightness may have a greater impact on image quality because it directly determines the brightness of the image and plays an important role in the visibility of details. In other scenarios, such as images with high color requirements, the weight of color saturation may be relatively high, and corresponding weights are allocated based on different needs.
[0073] S2. When the image quality does not meet the preset quality, detect the noise category of the original image, and based on the noise category, perform image quality enhancement on the original image to obtain an enhanced image, and when the enhanced image quality corresponding to the enhanced image meets the preset quality, obtain a first adjusted image.
[0074] The embodiment of the present invention can understand the specific reasons affecting the image quality by detecting the noise category of the original image when the image quality does not meet the preset quality, and then quickly solve the problem and improve the image quality. It should be noted that the image quality does not meet the preset quality means that the image quality does not meet the preset value. For example, in an indoor monitoring scene, the brightness evaluation value L is 0.7, the contrast evaluation value C is 0.8, the color saturation evaluation value S is 0.9, the clarity evaluation value D is 0.6, and the weights of each feature are respectively , then the image quality assessment index , according to the preset threshold value of 0.8, it is determined whether the image quality needs further adjustment. The noise categories include Gaussian noise, salt and pepper noise, and gamma noise.
[0075] As an embodiment of the present invention, the detection of the noise category of the original image includes: performing an original two-dimensional discrete Fourier transform on the original image to obtain frequency domain data about the original image, querying the amplitude spectrum and phase spectrum in the frequency domain data, analyzing the frequency domain characteristics of the original image based on the amplitude spectrum and the phase spectrum, and determining the noise category of the original image based on the frequency domain characteristics. The amplitude spectrum is used to describe the amplitude of each frequency component in the signal, and the phase spectrum is used to describe the phase information of each frequency component in the signal.
[0076] Optionally, the original image is subjected to an original two-dimensional discrete Fourier transform to obtain frequency domain data about the original image, which can be obtained by performing Fourier transform on an image processing library in a programming language (such as Python). The process of analyzing the frequency domain characteristics of the original image based on the amplitude spectrum and the phase spectrum is as follows: comparing the amplitude spectrum and the phase spectrum with the amplitude spectrum and the phase spectrum corresponding to different images to observe the frequency domain differences of the original image, and then analyzing the frequency domain changes of the original image to understand the frequency domain characteristics of the original image. The noise category of the original image is determined based on the frequency domain characteristics by analyzing the energy distribution characteristics in the frequency domain characteristics. If the energy distribution characteristics are uniform without obvious concentration or specific patterns, they are generally Gaussian noise. If the energy distribution characteristics are discrete high-frequency components, they may be salt and pepper noise.
[0077] Furthermore, the embodiment of the present invention performs image quality enhancement on the original image based on the noise category to obtain an enhanced image, which can improve image quality and make the photographing of the monitored area more accurate.
[0078] As an embodiment of the present invention, the image quality enhancement of the original image based on the noise category to obtain an enhanced image includes: adaptive denoising of the original image based on the noise category to obtain a denoised image, sharpening the denoised image to obtain a sharpened image, calculating a brightness distribution histogram of the sharpened image, adjusting the brightness of the sharpened image based on the calculation result of the brightness distribution histogram to obtain a brightness adjusted image, and color correcting the brightness adjusted image to obtain an enhanced image. The brightness distribution histogram refers to a distribution of the number of pixels at different brightness levels in the image.
[0079] Optionally, the adaptive denoising of the original image based on the noise category to obtain the denoised image refers to configuring the denoising method for the original image based on the noise category, such as Gaussian noise can use a mean filtering method, and salt and pepper noise can use a median filtering method. Optionally, the sharpening of the denoised image to obtain the sharpened image refers to a method for enhancing image details, which can be sharpened by a high-pass filter. The process of calculating the brightness distribution histogram of the sharpened image is: dividing the brightness value of the sharpened image into multiple intervals, and calculating the number of pixels in the multiple intervals, plotting the number of pixels in each interval in the histogram, the horizontal axis represents the brightness level, and the vertical axis represents the number of pixels. The brightness of the sharpened image is adjusted based on the calculation result of the brightness distribution histogram to obtain a brightness adjusted image through Adobe Photoshop tool adjusts contrast parameters for adjustment, and the color correction of the brightness adjusted image to obtain an enhanced image refers to adjusting the color of the image to make it more accurately reflect the color of the original scene, or to achieve a certain specific artistic effect. White balance adjustment can be used to correct the white points in the image to make it look closer to real white, color balance adjustment is used to adjust the intensity of different color channels in the image to change the overall color balance, and curve adjustment is used to adjust the tone curve of the image to change the color. The specific adjustment is made according to actual needs.
[0080] S3. Identify a detection target of the first adjusted image, construct an initial image frame of the first adjusted image based on the detection target, identify feature points of the detection target, and perform frame adjustment on the initial image frame based on the feature points to obtain a target frame.
[0081] The embodiment of the present invention can clarify the specific category of each target in the image by identifying the detection target of the first adjusted image, such as distinguishing whether it is a person, an animal or an object, and then continuously track the detected target in a continuous image sequence to understand its movement trajectory and behavior.
[0082] Optionally, the detection target of the first adjusted image can be identified by an R-CNN algorithm.
[0083] Furthermore, the embodiment of the present invention can obtain a detection tool for quickly locating a target by constructing an initial image frame of the first adjusted image based on the detection target.
[0084] As an embodiment of the present invention, constructing an initial image frame of the first adjusted image based on the detected target includes: calculating the target size of the detected target, constructing the frame size of the initial image based on the target size, identifying the position distribution information of the detected target in the first adjusted image, constructing a frame distribution strategy of the initial image based on the position distribution information, and constructing the initial image frame of the first adjusted image based on the frame size and the frame distribution strategy.
[0085] Optionally, the target size of the detection target is calculated by extracting the outline of the target object, and then calculating the size according to the proportional relationship between pixels and actual size. The size of the scanning frame is the same as the target size. For example, if the target size is 4a in length and 2a in width, the scanning frame size is a rectangle with a length of 4a and a width of 2a (a is a measurement unit, such as cm, which needs to be determined according to the actual observed target and image size). The position distribution information of identifying the detection target in the first adjusted image refers to the spatial distribution of the target in the image. For example, in a certain image, the target to be observed is distributed in the lower left corner, or is irregularly distributed. The scanning frame distribution strategy of the initial image is constructed based on the position distribution information, and a corresponding distribution is performed according to the distribution of the target. For example, if the target is distributed in the lower left corner of the image, the scanning frame is arranged in the lower left corner.
[0086] Furthermore, the embodiment of the present invention can track the changing target in real time by identifying the feature points of the detection target, so that the image can still maintain good stability and not be lost when it is rotated, scaled, or has brightness changes.
[0087] As an embodiment of the present invention, the identifying feature points of the detection target includes: constructing a scale space of the detection target, marking key points of the detection target in the scale space to obtain marked key points, assigning directions to the marked key points based on the marked key points, and determining feature points of the marked key points using the following formula to obtain feature points:
[0088] ;
[0089] in, Represents feature points, Indicates the image gradient value of the detected target corresponding to the first adjusted image, Indicates the direction of the marked key point, represents the feature image, , , , Indicates the origin position of the feature image in the scale space The nearest neighboring points.
[0090] Optionally, the scale space is obtained by performing Gaussian blurring of different scales on the first adjusted image corresponding to the detection target, and the key points of the detection target are marked in the scale space, and the marked key points can be detected by a SIFT algorithm. If the target to be detected is a person, the detection features of the person are input into the SIFT algorithm, and the SIFT algorithm is used to automatically mark the key points.
[0091] Furthermore, in another optional embodiment of the present invention, the direction assignment of the marked key points includes: calculating the domain gradient of the key points, constructing a direction histogram of the marked key points according to the gradient direction corresponding to the domain gradient, and taking the peak value in the direction histogram as the direction of the marked key points.
[0092] Furthermore, in the embodiment of the present invention, by adjusting the initial image frame based on the feature points to obtain the target frame, the frame can mark the target more accurately and the target is not easily lost.
[0093] As an embodiment of the present invention, the frame adjustment of the initial image frame based on the feature points to obtain the target frame includes: constructing a real bounding box of the first adjusted image based on the feature points, calculating the intersection and union ratio of the real bounding box and the initial image frame, and adjusting the frame of the initial image frame based on the intersection and union ratio to obtain the target frame. The real bounding box refers to a rectangular box in an image or video that is used to accurately mark the actual position and range of the target object.
[0094] Optionally, constructing the real bounding box of the first adjusted image based on the feature points can be constructed by a manual labeling method, calculating the intersection-and-union ratio of the real bounding box and the initial image frame refers to calculating the ratio of the intersection area of the two frames to the union area, and adjusting the initial image frame based on the intersection-and-union ratio to obtain the target frame by determining the matching relationship between the anchor frame and the real bounding box according to the size of the intersection-and-union ratio. If the value of the intersection-and-union ratio is 0.7, the initial image frame is used as the target frame. If the value of the intersection-and-union ratio is less than 0.7, the position of the initial image frame is adjusted until the intersection-and-union ratio reaches 0.7. It should be noted that the real bounding box is more accurate than the frame marking, but the operation steps are more complicated and it is not suitable for target tracking. The frame is convenient and fast and suitable for tracking, but its accuracy is slightly lower than that of the real bounding box. Therefore, this solution uses the real bounding box to adjust the accuracy of the initial image frame before starting.
[0095] S4. Track the detection target using the target frame to obtain a target trajectory, analyze a behavior pattern of the detection target based on the target trajectory, identify the clarity of the detection target when the behavior pattern is abnormal, and perform a second adjustment on the first adjustment image to obtain a second adjustment image when the clarity does not meet a preset requirement.
[0096] The embodiment of the present invention tracks the detection target by using the target trace frame to obtain a target trajectory, which can visualize the detection target and further present the movement of the target more intuitively, making it convenient for users to observe and analyze.
[0097] Optionally, the target tracking frame is used to track the detected target to obtain a target trajectory; the detected target is continuously image-marked by the target tracking frame to obtain a multi-frame marked target; the multi-frame marked targets are connected to obtain a target trajectory.
[0098] Furthermore, the embodiment of the present invention can understand the behavior pattern of the detection target by analyzing the behavior pattern of the detection target based on the target trajectory, such as walking route, stay time, activity range, etc. This is of great significance for studying animal behavior, human activities, traffic flow, etc.
[0099] As an embodiment of the present invention, the analyzing the behavior pattern of the detection target based on the target trajectory includes: extracting the behavior characteristics of the target trajectory, visually displaying the target trajectory based on the behavior characteristics to obtain a visualized trajectory, and performing behavior semantic analysis on the visualized trajectory to obtain a behavior pattern about the detection target.
[0100] Optionally, the behavioral features of the target trajectory can be extracted by a deep learning target detection algorithm (such as FasterR-CNN, YOLO, etc.), wherein the behavioral features include the length, speed, acceleration, turning angle, dwell time, etc. of the target trajectory; based on the behavioral features, the target trajectory is visualized to obtain a visualized trajectory; through virtual reality technology practice, the visualized trajectory is subjected to behavioral semantic analysis to obtain a behavioral pattern of the detected target; the trajectories are classified or grouped by using methods such as support vector machines, decision trees, and cluster analysis to discover different behavioral patterns, and then the target behavior pattern is explained in combination with domain knowledge or prior knowledge.
[0101] It should be noted that when the behavior pattern is abnormal, it may indicate the occurrence of an abnormal situation, such as intrusion, theft, traffic accident, etc., and then identify the clarity of the detection target to determine whether the recognition of the target detection object is clear.
[0102] Furthermore, the embodiment of the present invention performs a second adjustment on the first adjusted image when the clarity does not meet the preset requirements, and the second adjusted image can improve the image quality when the clarity of the target detection object is not enough, so as to improve the recognition of the captured image, and can be adjusted by adjusting parameters such as the resolution, frame rate and white balance of the monitoring equipment.
[0103] S5. Analyze the image adjustment level of the second adjusted image, and when the image adjustment level meets a preset level, use the second adjusted image as a target adjusted image.
[0104] The embodiment of the present invention can evaluate whether the quality of the processed image is improved or reduced compared with the original image by analyzing the image adjustment level of the second adjusted image, and analyze the quality loss of the image during compression and transmission.
[0105] As an embodiment of the present invention, the analyzing the image adjustment level of the second adjusted image includes: calculating the image mean square error of the second adjusted image using the following formula:
[0106] ;
[0107] in, represents the mean square error of the image, Indicates that the first adjusted image corresponding to the second adjusted image is at position The pixel value of Indicates that the second adjusted image is at position The pixel value of , H represents the height of the second adjusted image, Indicates the width of the second adjusted image,
[0108] Based on the mean square error, the peak signal-to-noise ratio of the second adjusted image is calculated:
[0109] ;
[0110] in, represents the peak signal-to-noise ratio, represents the mean square error of the image, represents the maximum pixel value of the second adjusted image,
[0111] An image adjustment level of the second adjusted image is determined based on the image mean square error and the peak signal-to-noise ratio.
[0112] Optionally, the image adjustment level of the second adjusted image is determined based on the image mean square error and the peak signal-to-noise ratio by comparing the image mean square error with a preset mean square error value. For example, if the preset mean square error value is 10, if the calculated image mean square error is less than or equal to 10, it can be considered to meet the preset value. The preset threshold corresponding to the peak signal-to-noise ratio is 5.0. If the calculated result is not less than 5.0, it indicates that the second adjusted image meets the image analysis requirements. If the image adjustment level does not meet the requirements, the image processing model corresponding to the second adjusted image (such as a deep learning model, which can be configured according to actual analysis requirements) is optimized for model adjustment until the second adjusted image meets the processing requirements.
[0113] Furthermore, the embodiment of the present invention can obtain a high-definition analysis image by using the second adjusted image as a target adjustment image when the image adjustment level meets a preset level, and can clearly observe or analyze the target in the image, thereby improving the high-definition video surveillance effect.
[0114] like Figure 2 , which is a functional module diagram of an image quality adaptive adjustment system in a high-definition video surveillance system provided by an embodiment of the present invention.
[0115] The image quality adaptive adjustment system 200 in a high-definition video surveillance system of the present invention can be installed in an electronic device. According to the functions to be implemented, the image quality adaptive adjustment system 200 in a high-definition video surveillance system may include an image quality analysis module 201, an image first adjustment module 202, a frame adjustment module 203, an image second adjustment module 204, and a target image acquisition module 205. The module of the present invention may also be referred to as a unit, which refers to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, and is stored in a memory of an electronic device.
[0116] In this embodiment, the functions of each module / unit are as follows:
[0117] The image quality analysis module 201 is used to obtain the original image in the high-definition video surveillance system, extract the basic feature information of the original image, and analyze the image quality of the original image based on the basic feature information, wherein the basic feature information includes brightness, contrast, color saturation and clarity;
[0118] The first image adjustment module 202 is configured to detect the noise category of the original image when the image quality does not meet the preset quality, enhance the image quality of the original image based on the noise category to obtain an enhanced image, and obtain a first adjusted image when the enhanced image quality corresponding to the enhanced image meets the preset quality;
[0119] The frame adjustment module 203 is used to identify the detection target of the first adjustment image, construct an initial image frame of the first adjustment image based on the detection target, identify the feature points of the detection target, and perform frame adjustment on the initial image frame based on the feature points to obtain a target frame;
[0120] The image second adjustment module 204 is used to track the detection target using the target frame to obtain a target trajectory, analyze the behavior pattern of the detection target based on the target trajectory, identify the clarity of the detection target when the behavior pattern is abnormal, and perform a second adjustment on the first adjustment image to obtain a second adjustment image when the clarity does not meet a preset requirement;
[0121] The target image acquisition module 205 is used to analyze the image adjustment level of the second adjusted image, and when the image adjustment level meets a preset level, use the second adjusted image as a target adjusted image.
[0122] In detail, the modules described in the image quality adaptive adjustment system 200 in the high-definition video surveillance system described in the embodiment of the present invention adopt the same technical means as the image quality adaptive adjustment method in the high-definition video surveillance system described in the accompanying drawings when used, and can produce the same technical effects, which will not be repeated here.
[0123] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.
[0124] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0125] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.
[0126] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0127] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.
[0128] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0129] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems stated in a system claim can also be implemented by one unit or system through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.
[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.
Claims
1. A method for adaptively adjusting image quality in a high-definition video surveillance system. It is characterized in that The method comprises: Acquire an original image in a high-definition video surveillance system, extract basic feature information of the original image, and analyze the image quality of the original image based on the basic feature information, wherein the basic feature information includes brightness, contrast, color saturation and clarity; When the image quality does not meet the preset quality, detecting the noise category of the original image, based on the noise category, performing image quality enhancement on the original image to obtain an enhanced image, and when the enhanced image quality corresponding to the enhanced image meets the preset quality, obtaining a first adjusted image; Identifying a detection target of the first adjusted image, constructing an initial image frame of the first adjusted image based on the detection target, identifying feature points of the detection target, and performing frame adjustment on the initial image frame based on the feature points to obtain a target frame, wherein, based on the noise category, performing image quality enhancement on the original image to obtain an enhanced image, comprises: Adaptively denoising the original image based on the noise category to obtain a denoised image; Performing sharpening processing on the denoised image to obtain a sharpened image; Calculating a brightness distribution histogram of the sharpened image; Performing brightness adjustment on the sharpened image based on the calculation result of the brightness distribution histogram to obtain a brightness adjusted image; Performing color correction on the brightness-adjusted image to obtain an enhanced image; The step of constructing an initial image frame of the first adjusted image based on the detected target includes: Calculating the target size of the detection target; Constructing a frame size of the initial image based on the target size; Identifying position distribution information of the detection target in the first adjusted image; Based on the position distribution information, construct a frame distribution strategy for the initial image; constructing an initial image frame of the first adjusted image based on the frame size and the frame distribution strategy; The identifying feature points of the detection target includes: Constructing a scale space of the detection target; Marking key points of the detection target in the scale space to obtain marked key points; Based on the marked key points, after the directions of the marked key points are assigned, the feature points of the marked key points are located using the following formula to obtain the feature points: Wherein, FP represents a feature point, K represents the image gradient value of the first adjusted image corresponding to the detection target, θ represents the direction of the marked key point, P represents a feature image, and P(A+1, B), P(A-1, B), P(A, B+1), and P(A, B-1) represent the neighboring points of the origin P(A, B) of the feature image in the scale space; The assigning directions to the marked key points includes: Calculating the domain gradient of the key point; Constructing a direction histogram of the marked key points according to the gradient direction corresponding to the domain gradient; Taking the peak value in the direction histogram as the direction of the marked key point; The step of adjusting the initial image frame based on the feature points to obtain a target frame includes: Based on the feature points, construct a real bounding box of the first adjusted image; Calculating the intersection-over-union ratio of the real bounding box and the initial image frame; Adjusting the initial image frame based on the intersection-and-union comparison to obtain a target frame; The target frame is used to track the detection target to obtain a target trajectory, and a behavior pattern of the detection target is analyzed based on the target trajectory. When the behavior pattern is abnormal, the clarity of the detection target is identified. When the clarity does not meet a preset requirement, a second adjustment is performed on the first adjustment image to obtain a second adjustment image, wherein the analysis of the behavior pattern of the detection target based on the target trajectory includes: Extracting behavioral features of the target trajectory; Based on the behavior characteristics, the target trajectory is visualized to obtain a visualized trajectory; Performing behavioral semantic analysis on the visualization trajectory to obtain a behavioral pattern of the detection target; An image adjustment level of the second adjustment image is analyzed, and when the image adjustment level meets a preset level, the second adjustment image is used as a target adjustment image.
2. The method for adaptively adjusting image quality in a high-definition video surveillance system according to claim 1, It is characterized in that The analyzing the image quality of the original image based on the basic feature information includes: Constructing weights corresponding to the basic feature information to obtain brightness evaluation weights, contrast weights, color saturation weights, and clarity weights; Based on the brightness evaluation weight, the contrast weight, the color saturation weight, and the clarity weight, the image quality of the original image is calculated in combination with the following formula: Q=w 1 ×L+w 2 ×C+w 3 ×S+w 4 ×D Where Q represents the image quality, L represents the brightness evaluation value corresponding to the basic feature information, C represents the contrast evaluation value corresponding to the basic feature information, S represents the color saturation evaluation value corresponding to the basic feature information, D represents the clarity evaluation value corresponding to the basic feature information, and w 1 represents the brightness evaluation weight, w 2 represents the contrast weight, w 3 represents the color saturation weight, w 4 Indicates the clarity weight.
3. The method for adaptively adjusting image quality in a high-definition video surveillance system according to claim 1, It is characterized in that The detecting the noise category of the original image comprises: Performing an original two-dimensional discrete Fourier transform on the original image to obtain frequency domain data about the original image; Querying the amplitude spectrum and phase spectrum in the frequency domain data; Analyzing frequency domain features of the original image based on the amplitude spectrum and the phase spectrum; The noise category of the original image is determined based on the frequency domain features.
4. An image quality adaptive adjustment system in a high-definition video surveillance system, It is characterized in that The method for adaptively adjusting image quality in a high-definition video surveillance system according to any one of claims 1 to 3 comprises: An image quality analysis module is used to obtain an original image in a high-definition video surveillance system, extract basic feature information of the original image, and analyze the image quality of the original image based on the basic feature information, wherein the basic feature information includes brightness, contrast, color saturation and clarity; a first image adjustment module, configured to detect a noise category of the original image when the image quality does not meet a preset quality, enhance the image quality of the original image based on the noise category to obtain an enhanced image, and obtain a first adjusted image when the enhanced image quality corresponding to the enhanced image meets the preset quality; A frame adjustment module is used to identify a detection target of the first adjustment image, construct an initial image frame of the first adjustment image based on the detection target, identify feature points of the detection target, and perform frame adjustment on the initial image frame based on the feature points to obtain a target frame, wherein, based on the noise category, image quality enhancement is performed on the original image to obtain an enhanced image, including: Adaptively denoising the original image based on the noise category to obtain a denoised image; Performing sharpening processing on the denoised image to obtain a sharpened image; Calculating a brightness distribution histogram of the sharpened image; Performing brightness adjustment on the sharpened image based on the calculation result of the brightness distribution histogram to obtain a brightness adjusted image; Performing color correction on the brightness-adjusted image to obtain an enhanced image; The step of constructing an initial image frame of the first adjusted image based on the detected target includes: Calculating the target size of the detection target; Constructing a frame size of the initial image based on the target size; Identifying position distribution information of the detection target in the first adjusted image; Based on the position distribution information, construct a frame distribution strategy for the initial image; constructing an initial image frame of the first adjusted image based on the frame size and the frame distribution strategy; The identifying feature points of the detection target includes: Constructing a scale space of the detection target; Marking key points of the detection target in the scale space to obtain marked key points; Based on the marked key points, after the directions of the marked key points are assigned, the feature points of the marked key points are located using the following formula to obtain the feature points: Wherein, FP represents a feature point, K represents the image gradient value of the first adjusted image corresponding to the detection target, θ represents the direction of the marked key point, P represents a feature image, and P(A+1, B), P(A-1, B), P(A, B+1), and P(A, B-1) represent the neighboring points of the origin P(A, B) of the feature image in the scale space; The assigning directions to the marked key points includes: Calculating the domain gradient of the key point; Constructing a direction histogram of the marked key points according to the gradient direction corresponding to the domain gradient; Taking the peak value in the direction histogram as the direction of the marked key point; The step of adjusting the initial image frame based on the feature points to obtain a target frame includes: Based on the feature points, construct a real bounding box of the first adjusted image; Calculating the intersection-over-union ratio of the real bounding box and the initial image frame; Adjusting the initial image frame based on the intersection-and-union comparison to obtain a target frame; The second image adjustment module is used to track the detection target using the target frame to obtain a target trajectory, analyze the behavior pattern of the detection target based on the target trajectory, identify the clarity of the detection target when the behavior pattern is abnormal, and perform a second adjustment on the first adjustment image to obtain a second adjustment image when the clarity does not meet the preset requirements, wherein the analysis of the behavior pattern of the detection target based on the target trajectory includes: Extracting behavioral features of the target trajectory; Based on the behavior characteristics, the target trajectory is visualized to obtain a visualized trajectory; Performing behavioral semantic analysis on the visualization trajectory to obtain a behavioral pattern of the detection target; The target image acquisition module is used to analyze the image adjustment level of the second adjustment image, and when the image adjustment level meets a preset level, use the second adjustment image as a target adjustment image.
