An oil depot safety management and monitoring system and method

Abstract

This application provides an oil depot safety management and monitoring system and method. The system parses the monitoring video stream of a target oil depot into multiple image frames, then separates the ambient light component from each image frame. Based on the fluctuation relationship between all ambient light components and the oil vapor concentration in the target oil depot, it determines the visual noise interference characteristics in the monitoring video stream, and then filters out distorted video frames from the monitoring video stream using these noise interference characteristics. It extracts the motion vector field of pixels between adjacent image frames in the monitoring video stream, and then determines the motion blur index of each image frame based on the motion vector field. Based on each motion blur index, it extracts the blur enhancement features of visual distortion in the monitoring video stream from the distorted video frames. Based on the blur enhancement features, it performs visual enhancement on the monitoring video stream to obtain the enhanced oil depot safety monitoring video. Using the scheme of this application, distorted areas in oil depot safety monitoring videos can be visually enhanced.

Description

Technical Field

[0001] This application relates to the field of video surveillance processing technology, and more specifically, to an oil depot safety management and monitoring system and method. Background Technology

[0002] With the rapid development of the petroleum storage and transportation industry, oil depots, as important sites for petroleum storage and transshipment, are facing increasing demands for safety management. Due to the highly flammable and explosive nature of petroleum products, any leaks, vapor diffusion, or improper operation may lead to major safety accidents. Therefore, real-time, efficient, and reliable monitoring of oil depot sites has become a core means of ensuring safe operation.

[0003] Oil depot safety management and monitoring refers to a comprehensive management approach that utilizes technologies such as video surveillance, environmental sensing, and intelligent analysis to conduct real-time and continuous safety status monitoring and risk warning for key areas within an oil depot, including oil storage areas, pipelines, and loading / unloading areas. This typically involves efficiently and reliably transmitting on-site video information from the oil depot to designated display terminals to intelligently identify and promptly intervene in abnormal behaviors such as fire hazards, oil and gas leaks, and personnel violations, ensuring safe operation in the high-risk environment of the oil depot. However, in existing oil depot safety management and monitoring systems, video surveillance is limited by complex environmental conditions (such as changes in lighting, oil vapor interference, and equipment vibration), resulting in a large amount of blurriness, distortion, and visual noise in the monitoring videos. This severely affects the reliability of timely identification of abnormal events and risk warnings, making it difficult for current oil depot safety management to make timely and accurate comprehensive judgments and rapid responses in the face of sudden fires, leaks, and illegal intrusions. This leads to frequent false alarms and missed alarms in oil depot safety management. Therefore, how to visually enhance distorted areas in oil depot safety monitoring videos has become a challenge for the industry. Summary of the Invention

[0004] This application provides an oil depot safety management and monitoring system and method, which can visually enhance distorted areas in oil depot safety monitoring videos.

[0005] In a first aspect, this application provides a method for enhancing oil depot monitoring video, comprising the following steps: The monitoring video stream of the target oil depot is parsed into multiple image frames, and then the ambient light component is separated from each image frame. Based on the relationship between all ambient light components and the fluctuation of oil vapor concentration in the target oil depot, the visually perceived noise interference characteristics in the monitoring video stream are determined, and then the distorted video frames in the safety monitoring and shooting process of the target oil depot are screened out from the monitoring video stream through the noise interference characteristics. The motion vector field of pixels between adjacent image frames in the monitoring video stream is extracted, and then the motion blur index of each image frame is determined based on the difference in gradient distribution of the image edges between the motion vector field and adjacent image frames. Based on various motion blur indices, extract the blur enhancement features of visual distortion in the surveillance video stream from the distorted video frames; The monitoring video stream is visually enhanced based on the fuzzy enhancement features to obtain an enhanced oil depot safety monitoring video.

[0006] In some embodiments, separating the ambient lighting component from each image frame specifically includes: Select an image frame as the selected image frame; Perform color space conversion on the selected image frame to obtain the color space of the selected image frame; Isolate the main channels of illumination effects from the color space; The ambient lighting component of a selected image frame is determined by the separated main channels; Continue to determine the ambient lighting components of the remaining image frames.

[0007] In some embodiments, determining the visually perceived noise interference characteristics in the monitoring video stream based on the relationship between all ambient light components and the fluctuation of oil vapor concentration in the target oil depot specifically includes: Obtain oil vapor concentration data in the target oil depot, and extract the fluctuation relationship of oil vapor concentration in the target oil depot from the oil vapor concentration data; The trend of light intensity variation in the monitoring video stream was determined by all ambient light components. Based on the fluctuation relationship and the changing trend of the light intensity, the visually perceived noise interference characteristics in the monitoring video stream are determined.

[0008] In some embodiments, filtering out distorted video frames from the monitoring video stream during the target oil depot safety monitoring process using the noise interference characteristics specifically includes: The noise impact degree in the monitoring video stream is determined by the noise interference characteristics. Based on the noise impact level, distorted video frames during the safety monitoring and filming process of the target oil depot are filtered out from the monitoring video stream.

[0009] In some embodiments, extracting the motion vector field of pixels between adjacent image frames in the monitoring video stream specifically includes: Select one image frame from the monitoring video stream as the selected image frame; Determine the adjacent frames of the selected image frame, and then combine the selected image frame with the adjacent frames to form the adjacent image frame corresponding to the selected image frame; Extract motion key points from the adjacent image frames; The pixel motion domain between adjacent image frames is determined by the displacement of motion key points in the adjacent image frames. Based on the pixel motion domain, a motion vector field is generated between pixels in adjacent image frames corresponding to the selected image frame; Continue to determine the motion vector field of pixels between adjacent image frames corresponding to the remaining image frames in the monitoring video stream.

[0010] In some embodiments, extracting the blur enhancement features of visual distortion in the surveillance video stream from the distorted video frames based on various motion blur indices specifically includes: Multiple blurred regions in the distorted video frame are extracted using various motion blur indices; Distortion analysis is performed on each fuzzy influence region to obtain the distortion loss of each fuzzy influence region; Based on all distortion losses, the blur enhancement features of visual distortion in the surveillance video stream are determined.

[0011] In some embodiments, visual enhancement of the monitoring video stream based on the fuzz enhancement features to obtain an enhanced oil depot safety monitoring video specifically includes: Based on the aforementioned blur enhancement features, a convolutional kernel is generated for visual distortion compensation of the surveillance video stream; The visual distortion compensation convolution kernel is used to visually enhance the monitoring video stream, resulting in an enhanced oil depot safety monitoring video.

[0012] Secondly, this application provides an oil depot safety management and monitoring system, including an oil depot monitoring video enhancement unit, wherein the oil depot monitoring video enhancement unit includes: The separation module is used to parse the monitoring video stream of the target oil depot into multiple image frames, and then separate the ambient light component from each image frame; The processing module is used to determine the visually perceived noise interference characteristics in the monitoring video stream based on the fluctuation relationship between all ambient light components and the oil vapor concentration in the target oil depot, and then filter out the distorted video frames in the target oil depot safety monitoring shooting process from the monitoring video stream through the noise interference characteristics. The processing module is also used to extract the motion vector field of pixels between adjacent image frames in the monitoring video stream, and then determine the motion blur index of each image frame based on the difference in gradient distribution of the motion vector field and the image edges between adjacent image frames. The processing module is also used to extract the blur enhancement features of visual distortion in the monitoring video stream from the distorted video frames according to each motion blur index; The execution module is used to perform visual enhancement on the monitoring video stream based on the fuzzy enhancement features to obtain the enhanced oil depot safety monitoring video.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described oil depot monitoring video enhancement method.

[0014] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned oil depot monitoring video enhancement method.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: In this application, the monitoring video stream of the target oil depot is parsed into multiple image frames, and then the ambient light component is separated from each image frame. Based on the fluctuation relationship between all ambient light components and the oil vapor concentration in the target oil depot, the visual noise interference characteristics in the monitoring video stream are determined. Then, distorted video frames from the target oil depot's safety monitoring process are selected from the monitoring video stream using these noise interference characteristics. The motion vector field of pixels between adjacent image frames in the monitoring video stream is extracted, and the motion blur index of each image frame is determined based on the difference in gradient distribution between the motion vector field and the image edges between adjacent image frames. Based on each motion blur index, blur enhancement features of visual distortion in the monitoring video stream are extracted from the distorted video frames. The monitoring video stream is then visually enhanced based on these blur enhancement features to obtain an enhanced oil depot safety monitoring video.

[0016] Therefore, in this application, firstly, distorted video frames during the target oil depot safety monitoring process are filtered out from the monitoring video stream using the noise interference features. The visually perceived noise interference features in the monitoring video stream effectively identify distorted video frames whose visual quality is degraded due to environmental factors during the target oil depot safety monitoring process, thus providing a basis for subsequent enhancement processing. Secondly, the motion blur index of each image frame is determined based on the difference in gradient distribution of image edges between the motion vector field and adjacent image frames. This accurately quantifies the degree of motion blur caused by factors such as camera shake and rapid object movement, providing data support for subsequent enhancement steps and ensuring targeted optimization of blur distortion to improve the clarity of the monitoring video. Then, based on each motion blur index, blur enhancement features of visual distortion in the monitoring video stream are extracted from the distorted video frames, enabling the development of optimization strategies for different types of blur regions. This method enables differentiated enhancement processing, reduces artifacts, and ensures that the enhanced surveillance video maintains a natural look, thus improving its practical value. Finally, based on the aforementioned fuzzy enhancement features, the surveillance video stream is visually enhanced to obtain an enhanced oil depot safety monitoring video. By adaptively optimizing distorted areas using the fuzzy enhancement features, the clarity of the surveillance video can be restored, effectively improving blurred areas. The enhanced oil depot safety monitoring video is then transmitted in real-time to the oil depot safety management and monitoring center for display on the terminal screen. This ensures that the enhanced surveillance video stream can be transmitted to the oil depot safety management and monitoring center in a short time, enabling the identification of abnormal events with optimized clarity, improving the reliability of oil depot safety management, and providing more accurate decision support for emergencies at the oil depot. In summary, this solution can visually enhance distorted areas in oil depot safety monitoring videos. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is an exemplary flowchart of an oil depot monitoring video enhancement method according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the determination of noise interference characteristics according to some embodiments of this application; Figure 3 This is an exemplary flowchart illustrating the extraction of blurred regions according to some embodiments of this application; Figure 4This is a schematic diagram of the structure of an oil depot monitoring video enhancement unit according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing a video enhancement method for oil depot monitoring, according to some embodiments of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] refer to Figure 1 The figure is an exemplary flowchart of an oil depot monitoring video enhancement method according to some embodiments of this application. The oil depot monitoring video enhancement method 100 mainly includes the following steps: In step 101, the monitoring video stream of the target oil depot is parsed into multiple image frames, and then the ambient light component is separated from each image frame.

[0021] In specific implementation, the monitoring video stream of the target oil depot can be parsed into multiple image frames in the following way: the monitoring video stream captured by multispectral cameras deployed in the oil depot's operating area can be obtained for safety monitoring of the target oil depot, and then the acquired monitoring video stream can be decoded into multiple consecutive image frames using a video decoding algorithm (such as H.264 or H.265 decoder). By parsing the monitoring video stream into multiple image frames, the continuous monitoring video stream can be converted into static image frames, which facilitates subsequent frame-by-frame analysis of the illumination and motion information in the safety monitoring area of ​​the target oil depot. Other methods can also be used to parse the monitoring video stream in other embodiments, which are not limited here.

[0022] It should be noted that the monitoring video stream in this application refers to video data captured during on-site safety monitoring of the target oil depot, and the image frame in this application refers to a static image frame in the monitoring video stream captured during safety monitoring of the target oil depot.

[0023] In some embodiments, separating the ambient lighting component from each image frame can be achieved using the following steps: Select an image frame as the selected image frame; Perform color space conversion on the selected image frame to obtain the color space of the selected image frame; Isolate the main channels of illumination effects from the color space; The ambient lighting component of a selected image frame is determined by the separated main channels; Continue to determine the ambient lighting components of the remaining image frames.

[0024] In practical implementation, color space conversion of the selected image frame can be achieved in the following way: The OpenCV `cvtColor()` function can be used to convert the selected image frame from RGB format to a color space more suitable for illumination analysis, thus separating luminance and color information, such as HSV (hue-saturation-luminance), LAB (luminance-chrominance), and YCrCb (luminance-chrominance signal). The luminance component (such as the V channel of HSV, the L channel of LAB, and the Y channel of YCrCb) mainly reflects the illumination effect, more clearly characterizing illumination features and reducing color interference. Separating the main channels representing illumination effects from the color space can be achieved by separating the single channel that best represents the illumination information (such as the V channel of HSV, the L channel of LAB, and the Y channel of YCrCb). The ambient illumination component of a selected image frame can be determined by separating the main channel as the primary channel for illumination influence. This can be achieved by: enhancing the contrast of the channel data in the separated main channel through adaptive histogram equalization, and denoising it using Gaussian blurring or bilateral filtering to obtain illumination data with reduced noise and enhanced local illumination features, thereby improving the stability of illumination information. Then, statistical analysis (such as global mean calculation, sliding window partitioning, and Gaussian mixture modeling) can be used to mathematically model the denoised and enhanced illumination data (i.e., a mathematical model of illumination influence, which can effectively separate the principal components of illumination from noise). The resulting result can then be used as the ambient illumination component of the selected image frame to improve the accuracy of illumination separation. Other methods can also be used in other embodiments, which are not limited here.

[0025] It should be noted that the ambient light component in this application represents an index of the illumination influence data in the image frame, and therefore can be used to reflect the degree and spatial distribution of illumination in the image frame; the main channel in this application is single-channel data characterizing the influence of ambient light; the color space in this application represents different image color representation methods in the image frame, and the channels therein are used to independently characterize illumination information.

[0026] In step 102, the visually perceived noise interference characteristics in the monitoring video stream are determined based on the relationship between all ambient light components and the fluctuation of oil vapor concentration in the target oil depot. Then, the distorted video frames in the target oil depot safety monitoring shooting process are filtered out from the monitoring video stream through the noise interference characteristics.

[0027] In some embodiments, reference Figure 2As shown, this figure is an exemplary flowchart for determining noise interference characteristics in some embodiments of this application. In this embodiment, determining the visually perceived noise interference characteristics in the monitoring video stream based on the fluctuation relationship between all ambient light components and the oil vapor concentration in the target oil depot can be achieved by the following steps: First, in step 1021, oil vapor concentration data in the target oil depot is obtained, and the fluctuation relationship of oil vapor concentration in the target oil depot is extracted from the oil vapor concentration data; Secondly, in step 1022, the trend of light intensity variation in the monitoring video stream is determined by all ambient light components; Finally, in step 1023, the visually perceived noise interference characteristics in the monitoring video stream are determined based on the fluctuation relationship and the changing trend of the light intensity.

[0028] In specific implementation, acquiring oil vapor concentration data in the target oil depot and extracting the fluctuation relationship of oil vapor concentration from the oil vapor concentration data can be achieved in the following way: Oil vapor concentration data in the target oil depot can be acquired from sensors or monitoring equipment (i.e., data acquisition equipment storing oil vapor concentration data at different times in the target oil depot environment), and a time series sequence of oil vapor concentration can be constructed by combining it with time tags. Data smoothing techniques such as moving average or wavelet analysis are used to filter the original oil vapor concentration data, removing occasional outliers. Time series modeling (such as polynomial fitting) is then used to identify the changing trend and periodic characteristics of oil vapor concentration in the target oil depot over time (such as the periodicity, abrupt change points, and slope changes of oil vapor concentration over time). This extracts the patterns reflecting the fluctuations in oil vapor concentration due to operational behavior, diurnal variations, or environmental factors. The obtained results are used as the fluctuation relationship of oil vapor concentration. This fluctuation relationship, such as "increases during the day and decreases at night" or "short-term surges during operational periods," ensures the accuracy of subsequent analysis of the visual impact of oil vapor. The characteristics of light intensity in the surveillance video stream are determined by analyzing all ambient light components (such as brightness and color temperature) to perform time-series analysis (such as weighted moving average or trend detection) on the changes in light intensity. The visually perceived noise interference characteristics in the surveillance video stream are determined based on the fluctuation relationship and the light intensity variation trend by fusing the fluctuation relationship with the light intensity variation trend using statistical modeling or machine learning methods (such as regression analysis or Gaussian mixture model). These noise interference characteristics manifest as image blurring, color distortion, or other visual interference in the surveillance video stream, thus identifying image distortion caused by oil vapor concentration. Other methods may be used in other embodiments, and are not limited here.

[0029] It should be noted that the fluctuation relationship of oil vapor concentration in this application represents the changing trend of oil vapor concentration in the target oil depot. The fluctuation relationship of oil vapor concentration will affect the visual quality of the monitoring screen of the target oil depot. For example, when the oil vapor concentration in the target oil depot is too high, it will easily lead to blurring and distortion of the monitoring video. The changing trend of light intensity in this application represents the fluctuation of light in the monitoring video stream. The visual perception noise interference characteristics in this application represent the visual perception noise generated by the monitoring video stream under the influence of oil vapor concentration.

[0030] In some embodiments, filtering out distorted video frames from the surveillance video stream during the safety monitoring and recording process of a target oil depot based on the noise interference characteristics can be achieved through the following steps: The noise impact degree in the monitoring video stream is determined by the noise interference characteristics. Based on the noise impact level, distorted video frames during the safety monitoring and filming process of the target oil depot are filtered out from the monitoring video stream.

[0031] In specific implementation, determining the noise impact degree in the monitoring video stream based on the noise interference characteristics can be achieved in the following way: the noise interference characteristics can be quantitatively analyzed using feature extraction and analysis methods (such as statistical methods or machine learning models), and the obtained quantitative results can be used as the noise impact degree in the monitoring video stream. Specifically, mean-variance analysis or model fitting can be used to identify the spatial distribution and temporal characteristics of noise in the monitoring video stream through the noise interference characteristics, and then the degree of noise interference on the image quality of the monitoring video stream can be quantified based on the identification results to obtain the noise impact degree; based on the noise impact degree, the distorted visuals in the target oil depot safety monitoring shooting process can be screened from the monitoring video stream. The frequency frame can be implemented in the following way: First, image quality assessment is performed on each image frame in the monitoring video stream using image quality assessment indicators such as pixel contrast and texture analysis. Second, a threshold for distorted image frames in the monitoring video stream can be preset based on prior experience or artificial intelligence according to the noise impact degree. Distorted image frames are usually characterized by low image quality, blurriness, or color distortion. Then, the threshold determination method can be used to filter out image frames in the monitoring video stream whose image quality assessment results are lower than the preset threshold. The set of all filtered image frames is taken as the distorted video frames in the target oil depot safety monitoring shooting process. Other methods can also be used in other embodiments, which are not limited here.

[0032] It should be noted that the noise impact degree in this application refers to the degree of interference of noise on the image quality of the monitoring video stream; the distorted video frame in this application refers to the set of image frames in the monitoring video stream that have visual distortion.

[0033] In step 103, the motion vector field of pixels between adjacent image frames in the monitoring video stream is extracted, and then the motion blur index of each image frame is determined based on the difference in gradient distribution of the motion vector field and the image edges between adjacent image frames.

[0034] In some embodiments, extracting the motion vector field of pixels between adjacent image frames in the monitoring video stream can be achieved using the following steps: Select one image frame from the monitoring video stream as the selected image frame; Determine the adjacent frames of the selected image frame, and then combine the selected image frame with the adjacent frames to form the adjacent image frame corresponding to the selected image frame; Extract motion key points from the adjacent image frames; The pixel motion domain between adjacent image frames is determined by the displacement of motion key points in the adjacent image frames. Based on the pixel motion domain, a motion vector field is generated between pixels in adjacent image frames corresponding to the selected image frame; Continue to determine the motion vector field of pixels between adjacent image frames corresponding to the remaining image frames in the monitoring video stream.

[0035] In specific implementation, determining the adjacent frames of a selected image frame can be achieved in the following way: the position of the selected image frame in the monitoring video stream can be determined by frame index calculation, thereby taking the subsequent frame of the selected image frame as the adjacent frame of the selected image frame, and then forming the adjacent image frames corresponding to the selected image frame with the adjacent frame; extracting motion key points in the adjacent image frames can be achieved in the following way: key point detection can be performed on the adjacent image frames using optical flow methods (such as dense optical flow Farneback or sparse optical flow Lucas-Kanade), thereby obtaining the motion key points in the adjacent image frames, that is, estimating motion by calculating the displacement of pixel points between consecutive frames to ensure the accuracy of subsequent motion estimation; determining the pixel motion domain between adjacent image frames by the displacement of motion key points in the adjacent image frames can be achieved in the following way: motion key point matching (such as FLANN matching) can be used. The displacement of motion keypoints in adjacent image frames is calculated using KNN matching or similar methods. This involves determining the corresponding positions of the motion keypoints in different image frames to calculate their motion displacements. The set of motion displacements of all motion keypoints is then used as the pixel motion domain between adjacent image frames to ensure fine-grained accuracy in motion estimation. The generation of the motion vector field between pixels in adjacent image frames corresponding to the selected image frame can be achieved by using optical flow interpolation methods (such as Horn-Schunck global optical flow or Brox optical flow) to perform vector calculations on the data in the pixel motion domain. This yields the motion vector field between pixels in the selected image frame and adjacent frames, ensuring the globality of motion information. Other methods can also be used in other embodiments, which are not limited here.

[0036] It should be noted that, in this application, adjacent image frames refer to adjacent image frames in a surveillance video stream; motion keypoints refer to feature points in adjacent image frames used for motion estimation; pixel motion domains refer to the set of pixel-level displacement distributions in adjacent image frames, which can be used to describe fine-grained information about motion trajectories in adjacent image frames; and motion vector fields refer to the motion direction and velocity of different regions in two image frames corresponding to adjacent image frames.

[0037] In some embodiments, determining the motion blur index of each image frame based on the difference in gradient distribution of image edges between the motion vector field and adjacent image frames can be achieved using the following steps: Select an image frame as the selected image frame, and obtain the adjacent image frames corresponding to the selected image frame; Determine the gradient distribution differences of image edges between adjacent image frames; The sharpness attenuation of adjacent image frames is determined by the difference in gradient distribution. The motion blur index of the selected image frame is determined by combining the sharpness attenuation amount with the motion vector field. Continue to determine the motion blur index of the remaining image frames.

[0038] In specific implementation, determining the gradient distribution difference of image edges between adjacent image frames can be achieved in the following way: edge detection can be performed on the adjacent image frames using the Sobel operator, Laplacian operator, or Canny edge detection, and then gradient calculation can be performed on the edge images to obtain the gradient distribution of the two image frames in the adjacent image frames. The gradient changes between the two image frames are then compared to determine the gradient distribution difference of image edges between the adjacent image frames, thereby capturing the blurred areas caused by motion in the adjacent image frames. Determining the sharpness attenuation of adjacent image frames based on the gradient distribution difference can be achieved in the following way: an evaluation algorithm (e.g., genetic algorithm, ensemble learning, and reinforcement learning) is used to evaluate the degree of visual change (i.e., the degree of sharpness reduction) caused by motion between adjacent image frames based on the gradient distribution difference. The obtained evaluation value is used as the sharpness attenuation of the adjacent image frames. The greater the gradient distribution difference of image edges between adjacent image frames, the stronger the image blur in the adjacent image frames, indicating a greater degree of sharpness reduction caused by motion in the adjacent image frames, i.e., sharpness attenuation. The higher the value, the higher the motion blur index of the selected image frame can be determined by combining the sharpness attenuation with the motion vector field. This can be achieved by using a machine learning model (such as support vector regression or a neural network) to take the displacement information (such as motion direction and speed) in the motion vector field and the sharpness attenuation as input features. The machine learning model then uses historical experimental experience to predict the degree of motion blur within the selected image frame, quantifies and normalizes the prediction results, and uses the normalized output as the motion blur index of the selected image frame. Here, the motion vector field reflects the motion direction and speed of objects in the image frame, while the sharpness attenuation reflects the degree of sharpness reduction caused by motion. Therefore, combining the two can more accurately quantify the degree of motion blur of the image frame. In addition, historical experimental data is usually used to train the machine learning model to ensure that the model can accurately predict the motion blur index. During the training process, model parameters need to be adjusted, such as the kernel function selection of support vector regression (SVR), the number of layers and neurons in the neural network, etc. Other methods can also be used to determine the motion blur index in other embodiments, which are not limited here.

[0039] It should be noted that the motion blur index in this application represents the severity of motion blur within an image frame in a surveillance video stream; the gradient distribution difference in this application represents an index for judging the sharpness attenuation of image edges in adjacent image frames, used to analyze visual changes between adjacent image frames; and the sharpness attenuation amount in this application represents the degree of sharpness reduction caused by motion in adjacent image frames.

[0040] In step 104, blur enhancement features of visual distortion in the monitoring video stream are extracted from the distorted video frames according to each motion blur index.

[0041] In some embodiments, extracting the blur enhancement features of visual distortion in the surveillance video stream from the distorted video frames based on various motion blur indices can be achieved using the following steps: Multiple blurred regions in the distorted video frame are extracted using various motion blur indices; Distortion analysis is performed on each fuzzy influence region to obtain the distortion loss of each fuzzy influence region; Based on all distortion losses, the blur enhancement features of visual distortion in the surveillance video stream are determined.

[0042] For specific implementation, refer to Figure 3 As shown, this figure is an exemplary flowchart for determining and extracting blurred regions in some embodiments of this application. Extracting multiple blurred regions in the distorted video frame from various motion blur indices can be achieved in the following way: First, a blur threshold can be preset based on all motion blur indices. For example, the average value of all motion blur indices can be preset as the blur threshold, and it can be adjusted according to the specific application scenario to ensure that the preset blur threshold can accurately distinguish blurred regions from normal regions. Then, image frames in the distorted video frame with motion blur indices higher than the blur threshold are all considered as blurred regions. Through this step, the obviously blurred regions in the distorted video frame are distinguished from the normal regions. The regions are separated to focus on those areas that affect video quality. Distortion analysis is performed on each blurred region to obtain the distortion loss of each blurred region. This can be achieved by using image quality assessment algorithms (such as PSNR (Peak Signal-to-Noise Ratio) or SSIM (Structural Similarity Index)) to perform distortion analysis on each blurred region. Based on the comparison between each blurred region and the original image, the distortion loss of each blurred region is calculated, which can effectively assess the degree of image quality loss in the blurred regions and obtain a quantitative value of the distortion loss of each blurred region. Other methods can also be used to determine this in other embodiments, which are not limited here.

[0043] In specific implementation, determining the blur enhancement features of visual distortion in the monitoring video stream based on all distortion losses can be achieved in the following way: The distortion loss of each blurred influence region can be aggregated into a single blur enhancement feature using a weighted average method or a weighted fusion method. The weights need to be reasonably allocated according to the importance of each blurred influence region to ensure the accuracy of the final feature. This process uses the weighted value of the distortion loss of each blurred influence region to determine the contribution of each blurred influence region to the final blur enhancement feature, thereby extracting the visual distortion features in the monitoring video stream. Then, machine learning algorithms are used to enhance the extracted visual distortion features based on historical experimental data and experience. Finally, the enhanced feature vector is used as the blur enhancement feature of visual distortion in the monitoring video stream. This blur enhancement feature contains the enhanced visual distortion information in the monitoring video stream, providing a basis for improving the monitoring video quality of the target oil depot. Other methods can also be used in other embodiments, which are not limited here.

[0044] It should be noted that the blurred affected region in this application refers to the region in the distorted video frame where visual distortion is obvious due to motion blur; the distortion loss in this application refers to the quantified value of the image quality degradation caused by motion blur; and the blur enhancement feature in this application refers to the feature vector that enhances the visual distortion features in the surveillance video stream. Therefore, the blur enhancement feature is used for subsequent visual enhancement processing of the surveillance video stream.

[0045] In step 105, the monitoring video stream is visually enhanced based on the fuzzy enhancement feature to obtain the enhanced oil depot safety monitoring video.

[0046] In some embodiments, visual enhancement of the monitoring video stream based on the fuzzy enhancement features to obtain the enhanced oil depot safety monitoring video can be achieved through the following steps: Based on the aforementioned blur enhancement features, a convolutional kernel is generated for visual distortion compensation of the surveillance video stream; The visual distortion compensation convolution kernel is used to visually enhance the monitoring video stream, resulting in an enhanced oil depot safety monitoring video.

[0047] In specific implementation, the convolutional kernel for visual distortion compensation in the surveillance video stream based on the blur enhancement features can be implemented in the following way: A convolutional kernel for adaptive visual distortion compensation in the surveillance video stream can be designed based on the blur enhancement features using convolutional neural networks or filter design methods in image processing (such as Gaussian blur kernels, sharpening kernels, etc.). Specifically, a series of weights can be generated based on the enhanced visual distortion information in the blur enhancement features, and a convolutional kernel for compensating for image distortion in the surveillance video stream can be constructed using these weights. The design of this convolutional kernel can also be trained using optimization algorithms (such as backpropagation) to maximize the compensation for blur and distortion in the surveillance video stream; the convolutional kernel for visual distortion compensation... Visual enhancement of the monitoring video stream to obtain the enhanced oil depot safety monitoring video can be achieved in the following way: Visual enhancement processing of the monitoring video stream is performed using convolution operations and the convolution kernel for visual distortion compensation. Specifically, the convolution kernel is applied frame-by-frame to each image frame in the monitoring video stream, thereby effectively compensating for visual distortion areas in the monitoring video stream to improve image clarity and suppress noise. This process enhances details in the monitoring video stream, reduces the impact of noise and motion blur, and restores the clarity of the video images in the monitoring video stream, ultimately obtaining the visually enhanced oil depot safety monitoring video, which improves the clarity and detail of the oil depot safety monitoring video in terms of visual effect.

[0048] It should be noted that the convolution kernel for visual distortion compensation in this application is used to compensate for visual distortion in surveillance video streams.

[0049] In some embodiments, the enhanced oil depot safety monitoring video can be transmitted in real time to the oil depot safety management and monitoring center for display on the terminal screen. That is, the enhanced oil depot safety monitoring video is transmitted in real time to the oil depot safety management and monitoring center via network protocol, and then the visually enhanced monitoring video is displayed in real time on the terminal device of the oil depot safety management center to improve the safety monitoring capability of the target oil depot.

[0050] In another aspect, in some embodiments, this application provides an oil depot safety management and monitoring system, which further includes an oil depot monitoring video enhancement unit, as referenced. Figure 4 The figure is a schematic diagram of the structure of an oil depot monitoring video enhancement unit 400 according to some embodiments of this application. The oil depot monitoring video enhancement unit 400 includes: a separation module 401, a processing module 402, and an execution module 403, which are described below: The separation module 401 in this application is mainly used to parse the monitoring video stream of the target oil depot into multiple image frames, and then separate the ambient light component from each image frame. Processing module 402 in this application is mainly used to determine the visually perceived noise interference characteristics in the monitoring video stream based on the fluctuation relationship of all ambient light components and oil vapor concentration in the target oil depot, and then filter out the distorted video frames in the target oil depot safety monitoring shooting process from the monitoring video stream through the noise interference characteristics. The processing module 402 described in this application is further used to extract the motion vector field of pixels between adjacent image frames in the monitoring video stream, and then determine the motion blur index of each image frame based on the difference in gradient distribution of the motion vector field and the image edges between adjacent image frames. The processing module 402 described in this application is further configured to extract the blur enhancement features of visual distortion in the monitoring video stream from the distorted video frame according to each motion blur index; The execution module 403 in this application is mainly used to perform visual enhancement on the monitoring video stream based on the fuzzy enhancement features to obtain the enhanced oil depot safety monitoring video.

[0051] The foregoing has detailed examples of the oil depot safety management and monitoring system and method provided in the embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0052] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described oil depot monitoring video enhancement method.

[0053] In some embodiments, reference Figure 5 The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the computer device implementing the oil depot monitoring video enhancement method of this application. The oil depot monitoring video enhancement method in the above embodiments can be achieved through… Figure 5 The computer device 500 shown is used to implement this, and the computer device 500 includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device 500 may be a terminal device, a server or a chip.

[0054] The processor 501 can be a general-purpose processor or a special-purpose processor. For example, the processor 501 can be a central processing unit (CPU). The CPU can be used to control the computer device 500, execute software programs, and process data from the software programs. The computer device 500 may also include a communication unit 505 for inputting (receiving) and outputting (transmitting) signals.

[0055] For example, computer device 500 may be a chip, communication unit 505 may be the input and / or output circuit of the chip, or communication unit 505 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.

[0056] For example, computer device 500 may be a terminal device or a server, and communication unit 505 may be a transceiver of the terminal device or the server, or communication unit 505 may be a transceiver circuit of the terminal device or the server.

[0057] The computer device 500 may include one or more memories 502 storing a program 504. The program 504 can be executed by a processor 501 to generate instructions 503, causing the processor 501 to perform the methods described in the above method embodiments according to the instructions 503. Optionally, the memory 502 may also store data (such as a target audit model). Optionally, the processor 501 may also read data stored in the memory 502, which may be stored at the same storage address as the program 504, or the data may be stored at a different storage address than the program 504.

[0058] The processor 501 and memory 502 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.

[0059] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 501. The processor 501 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.

[0060] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0061] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described oil depot monitoring video enhancement method.

[0062] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0063] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for enhancing oil depot surveillance video, used to visually enhance surveillance video in an oil depot safety management monitoring system, characterized in that, The method includes the following steps: The monitoring video stream of the target oil depot is parsed into multiple image frames, and then the ambient light component is separated from each image frame. Based on the relationship between all ambient light components and the fluctuation of oil vapor concentration in the target oil depot, the visually perceived noise interference characteristics in the monitoring video stream are determined, and then the distorted video frames in the safety monitoring and shooting process of the target oil depot are screened out from the monitoring video stream through the noise interference characteristics. The motion vector field of pixels between adjacent image frames in the monitoring video stream is extracted, and then the motion blur index of each image frame is determined based on the difference in gradient distribution of the image edges between the motion vector field and adjacent image frames. Based on various motion blur indices, extract the blur enhancement features of visual distortion in the surveillance video stream from the distorted video frames; The monitoring video stream is visually enhanced based on the fuzzy enhancement features to obtain an enhanced oil depot safety monitoring video.

2. The method as described in claim 1, characterized in that, Separating the ambient lighting component from each image frame specifically includes: Select an image frame as the selected image frame; Perform color space conversion on the selected image frame to obtain the color space of the selected image frame; Isolate the main channels of illumination effects from the color space; The ambient lighting component of a selected image frame is determined by the separated main channels; Continue to determine the ambient lighting components of the remaining image frames.

3. The method as described in claim 1, characterized in that, Based on the relationship between all ambient light components and the fluctuation of oil vapor concentration in the target oil depot, the visually perceived noise interference characteristics in the monitoring video stream are determined, specifically including: Obtain oil vapor concentration data in the target oil depot, and extract the fluctuation relationship of oil vapor concentration in the target oil depot from the oil vapor concentration data; The trend of light intensity variation in the surveillance video stream was determined by all ambient light components. Based on the fluctuation relationship and the changing trend of the light intensity, the visually perceived noise interference characteristics in the monitoring video stream are determined.

4. The method as described in claim 1, characterized in that, Filtering out distorted video frames from the surveillance video stream using the noise interference characteristics specifically includes: The noise impact degree in the monitoring video stream is determined by the noise interference characteristics. Based on the noise impact level, distorted video frames during the safety monitoring and filming process of the target oil depot are filtered out from the monitoring video stream.

5. The method as described in claim 1, characterized in that, Extracting the motion vector field of pixels between adjacent image frames in the surveillance video stream specifically includes: Select one image frame from the monitoring video stream as the selected image frame; Determine the adjacent frames of the selected image frame, and then combine the selected image frame with the adjacent frames to form the adjacent image frame corresponding to the selected image frame; Extract motion key points from the adjacent image frames; The pixel motion domain between adjacent image frames is determined by the displacement of motion key points in the adjacent image frames. Based on the pixel motion domain, a motion vector field is generated between pixels in adjacent image frames corresponding to the selected image frame; Continue to determine the motion vector field of pixels between adjacent image frames corresponding to the remaining image frames in the monitoring video stream.

6. The method as described in claim 1, characterized in that, Extracting blur enhancement features for visual distortion in the surveillance video stream from the distorted video frames based on various motion blur indices specifically includes: Multiple blurred regions in the distorted video frame are extracted using various motion blur indices; Distortion analysis is performed on each fuzzy influence region to obtain the distortion loss of each fuzzy influence region; Based on all distortion losses, the blur enhancement features of visual distortion in the surveillance video stream are determined.

7. The method as described in claim 1, characterized in that, Visual enhancement of the monitoring video stream based on the aforementioned fuzzy enhancement features yields the enhanced oil depot safety monitoring video, specifically including: Based on the aforementioned blur enhancement features, a convolutional kernel is generated for visual distortion compensation of the surveillance video stream; The visual distortion compensation convolution kernel is used to visually enhance the monitoring video stream, resulting in an enhanced oil depot safety monitoring video.

8. An oil depot safety management and monitoring system, the system comprising an oil depot monitoring video enhancement unit, characterized in that, The oil depot monitoring video enhancement unit includes: The separation module is used to parse the monitoring video stream of the target oil depot into multiple image frames, and then separate the ambient light component from each image frame; The processing module is used to determine the visually perceived noise interference characteristics in the monitoring video stream based on the fluctuation relationship between all ambient light components and the oil vapor concentration in the target oil depot, and then filter out the distorted video frames in the target oil depot safety monitoring shooting process from the monitoring video stream through the noise interference characteristics. The processing module is also used to extract the motion vector field of pixels between adjacent image frames in the monitoring video stream, and then determine the motion blur index of each image frame based on the difference in gradient distribution of the motion vector field and the image edges between adjacent image frames. The processing module is also used to extract the blur enhancement features of visual distortion in the monitoring video stream from the distorted video frames according to each motion blur index; The execution module is used to perform visual enhancement on the monitoring video stream based on the fuzzy enhancement features to obtain the enhanced oil depot safety monitoring video.

9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, causing the computer device to perform the oil depot monitoring video enhancement method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the oil depot monitoring video enhancement method as described in any one of claims 1 to 7.

Images