71 results about "Fuzzy graph" patented technology

The invention discloses a fault diagnosis method and system based on a multi-source data association rule and a graph neural network. The method comprises the steps of extracting high-frequency operation data and low-frequency time sequence state data based on historical data, and establishing an equipment operation feature set; an Apriori algorithm is utilized to screen correlation characteristics to calculate a correlation relation, and a fault symptom set is constructed; and taking the association relationship of the features as an adjacent matrix embedded graph neural network, and training the constructed fuzzy graph neural network based on historical data to obtain a fault diagnosis model. The system comprises a data acquisition module, a preprocessing module, a feature extraction module, a feature screening module, a feature association relationship analysis module, a fault diagnosis model training module, a fault diagnosis module and a database storage module, and can perform multi-source data fusion analysis and training and updating of a fault diagnosis model. According to the method, the fault diagnosis model is constructed by combining multi-source data fusion, feature extraction, association relationship mining and the fuzzy graph neural network, so that more accurate and efficient fault diagnosis is realized.

The invention relates to a high-precision three-dimensional Gaussian reconstruction method based on fuzzy perception and joint optimization, and belongs to the field of computer vision and graphic images. The method comprises the following steps: firstly, constructing a fuzzy sensing module, accurately identifying a fuzzy type and an area of an input multi-view fuzzy image, and generating a corresponding fuzzy mask; performing three-dimensional Gaussian representation based on the input multi-view blurred image, estimating internal and external parameters of a camera and sparse point cloud data of a scene, and obtaining initial three-dimensional Gaussian scene representation; and then performing joint optimization on a Gaussian ellipsoid in the initial three-dimensional Gaussian scene representation in combination with an input multi-view blurred image and a blurred mask to generate a clear three-dimensional scene representation. And finally, performing three-dimensional Gaussian reconstruction rendering by using the optimized three-dimensional Gaussian scene representation, and outputting a high-precision three-dimensional Gaussian reconstruction result and a clear rendered image. The method can effectively improve the three-dimensional reconstruction precision and rendering quality under the fuzzy input condition, has high adaptability and stability, and can be applied to the fields of image deblurring, virtual reality and the like.

The invention relates to the technical field of computer vision, and discloses a robot image recognition system based on a convolutional neural network, and the system comprises a dynamic retina blur sensing module which is used for estimating a pixel-level dynamic blur kernel parameter from an input blur image; the differentiable physical deblurring layer is used for performing feature reconstruction on the blurred image based on the dynamic blurring kernel parameters to generate deblurred image features; and the fuzzy robust CNN backbone network dynamically adjusts the voidage of a convolutional layer according to the dynamic fuzzy kernel parameters so as to adapt to the local fuzzy intensity. The problem of parameter distortion of a data-driven model is solved through an optical flow constraint fuzzy perception module, dilated convolution is dynamically adjusted based on fuzzy intensity to expand a high fuzzy region receptive field, an end-to-end joint training framework is constructed to realize classification error reverse optimization fuzzy kernel parameters, and a two-stage curriculum learning strategy is combined to improve the robustness of the system. The physical model is solidified first and then global optimization is carried out, and the bottleneck of traditional staged training is broken through.

The invention relates to the field of track detection, in particular to a vision-based rapid identification method for a target object in a track area, which comprises a target object image sampling step, a fuzzy database construction step, a sampled image calibration step and a target object rapid identification step. The problems of image blurring, multi-source errors and real-time performance under high-speed movement are systematically solved; the beneficial effects of efficiently coping with image blurring and dynamic interference, improving system robustness and detection coverage rate through a multi-camera cooperative calibration mechanism, adapting to a complex operation environment through a dynamic error elimination strategy, and improving detection reliability and safety are achieved.

The invention discloses a toothbrush defect detection method and a sorting device, and belongs to the technical field of image processing, and the method comprises the following steps: establishing a feature space containing four-dimensional core parameters, forming a dynamic feature database, setting a difference identification method, constructing a basic defect model, monitoring the change condition in the production process in real time, and finely adjusting the basic defect model, the feature change is adapted; the method comprises the following steps: acquiring an original image sequence of a to-be-detected toothbrush, setting a blurring restoration method, and performing sharpening processing on a blurred image, thereby solving the problem of defect missing detection caused by image blurring in a dynamic shooting process, determining a key detection area of the to-be-detected toothbrush, facilitating accurate positioning of a defect area, reducing missing detection risk, setting a defect detection method, and improving the detection accuracy of the to-be-detected toothbrush. And performing defect detection on the key detection area by utilizing the finely-adjusted defect model, and grading the toothbrush to be detected by combining a grading rule.

The invention provides an image visual identification processing method for foundation pit monitoring, and belongs to the technical field of data processing, and the method comprises the steps: 1, carrying out the multi-modal image and motion information collection of a foundation pit; 2, the background server carries out multi-modal feature fusion and blurred image screening on the foundation pit image; step 3, the background server performs rigid-non-rigid structure classification and feature enhancement based on the multi-modal feature map; 4, the background server performs dynamic fuzzy kernel construction and adaptive deblurring on the rigid structure area and the non-rigid structure area; and 5, based on the complete deblurred image, fine crack identification of the foundation pit is carried out. Through four core steps of multi-modal feature fusion extraction, dynamic fuzzy kernel construction, non-rigid structure enhancement and refined crack identification, efficient deblurring and accurate crack identification of a foundation pit image are realized.

The invention relates to an implicit diffusion super-resolution assisted remote sensing image small target detection and identification method. The method comprises the following steps: acquiring a training data set; obtaining a clear-fuzzy sample pair; and introducing the target classification model and the implicit diffusion model into the target detection and recognition model to obtain a small target detection and recognition model. The recognition method is used for effectively supplementing feature information when small target detection and recognition are carried out, so that the accuracy of small target detection and recognition is improved. According to the invention, aiming at a difficult point with less effective information in small target detection and identification, a low-resolution blurred image is gradually recovered to a high-resolution clear image by means of an implicit diffusion model based on deep learning, super-division of an image area where a small target is located is realized, and the definition of the area image is improved, so that feature information is enriched, and the detection and identification efficiency is improved. The problems of insufficient feature information, unclear target boundary, susceptibility to noise interference and the like in small target detection and recognition are relieved, and the overall effect of remote sensing image target detection and recognition is finally improved.

The invention discloses a dynamic scene deblurring method and system based on physical information adversarial learning, and the method comprises the steps: obtaining original blurred image data, carrying out the preprocessing of the original blurred image data, and obtaining a preliminary deblurred image; inputting the preliminary deblurred image into an initial dynamic scene deblurring model for training to obtain a dynamic scene deblurring model; the initial dynamic scene deblurring model comprises a generator network, an optical flow estimation network, a three-stage progressive training strategy and a multi-scale discriminator network, the generator network is used for mapping an initial deblurred image into a deblurred image, and the optical flow estimation network is used for estimating a motion field and calculating optical flow consistency loss; the three-stage progressive training strategy is used for carrying out three-stage training on the initial dynamic scene fuzzy model in sequence, and the multi-scale discriminator network is used for judging image authenticity on different scales; and inputting a to-be-processed blurred image into the dynamic scene deblurring model to obtain a blurred image. According to the invention, the deblurring effect is improved.

The invention discloses a fuzzy type adaptive image restoration method, which belongs to the field of image processing and computer vision, and comprises the following steps: carrying out frame extraction processing on an input video stream, and preprocessing extracted image frames; based on the preprocessed image, a comprehensive fuzzy score is calculated through a multi-feature fuzzy evaluation mechanism; judging whether the image frame is a fuzzy frame according to the comprehensive fuzzy score; if the image frame is a fuzzy frame, judging the fuzzy type of the image frame through a preset rule according to a plurality of evaluation scores in a multi-feature fuzzy evaluation mechanism; selecting a corresponding restoration strategy from a plurality of preset restoration strategies to restore the image frame, and generating a restored image; performing quality evaluation on the repaired image, and if an evaluation result does not meet a preset condition, adjusting a repairing strategy and repairing again; and re-inserting the repaired image meeting the preset condition into the input video stream, and outputting the repaired video. According to the invention, efficient identification and accurate restoration of the blurred image frame are realized.

According to the high-speed fuzzy license plate character recognition method based on deep learning, a deep convolutional generative adversarial network BLPDCGAN generates a high-quality fuzzy image similar to a real fuzzy license plate, a data set is expanded, and the model generalization ability is improved. The generation module introduces a noise image to enhance the blurring effect, and optimizes the quality of the generated blurred image through MSE loss and adversarial loss. The blurred license plate character recognition module uses a Unet architecture for reference, integrates a CBAM attention mechanism and multi-scale feature fusion, enables the model to pay more attention to character features rather than noise, and improves the feature extraction capability of blurred images. The feature enhancement module further enhances the structure and texture extraction of the fuzzy characters, and increases the robustness and recognition precision of the model to the characters in the fuzzy scene. According to the multi-scale feature fusion strategy, features with different resolutions are aggregated, so that the model can capture key feature details of a fuzzy region and optimize the overall recognition effect.

The application discloses a kind of unmanned aerial vehicle image enhancement method and system.Therein, the method includes: obtaining first fuzzy image and its corresponding semantic prompt text;The structural feature extraction is carried out to first fuzzy image, and the semantic feature extraction is carried out to semantic prompt text, to obtain prior information;Prior information is injected into denoising model using cross attention mechanism;First fuzzy image is input into the denoising model injected with prior information to carry out multiple rounds of training and obtain target denoising model;Second fuzzy image is obtained, which is subjected to frequency domain transformation and input into frequency domain residual error model for multiple rounds of training to obtain target frequency domain residual error model;The fuzzy image to be detected is input into target denoising model for denoising to obtain spatial clear image;The fuzzy image to be detected is input into target frequency domain residual error model for prediction to obtain frequency domain clear image;Spatial clear image and frequency domain clear image are fused to obtain high-resolution image.The method can obtain higher quality clear image.

The invention relates to the technical field of three-dimensional reconstruction, and discloses a three-dimensional scene reconstruction method and device, equipment, a storage medium and a program product, and the method comprises the steps: carrying out the initialization of a three-dimensional Gaussian ellipsoid based on an obtained fuzzy image sequence and an event sequence, and obtaining an initial Gaussian ellipsoid; constructing a loss function based on a clear picture sequence obtained by rendering the fuzzy picture sequence through the initial Gaussian ellipsoid and geometric constraints of an event-ungenerated area corresponding to the fuzzy picture sequence and the event sequence; taking the minimum value of the loss function as an optimization target, and adjusting the parameter information of the initial Gaussian ellipsoid to obtain a target Gaussian ellipsoid; and performing three-dimensional scene reconstruction based on the target Gaussian ellipsoid to obtain a target three-dimensional scene. Therefore, three-dimensional reconstruction is carried out in combination with a fuzzy picture sequence obtained by a traditional camera and an event sequence obtained by an event camera, the reconstruction effect of a three-dimensional scene is improved, geometric constraint is added to guide and adjust a Gaussian ellipsoid, and the reconstruction effect of the three-dimensional scene is further improved.

The application belongs to the technical field of simultaneous localization and mapping, and particularly relates to a SLAM algorithm based on feature reinforcement and motion judgment in a dynamic scene, which applies a feature reinforcement instance segmentation network FENET and comprises the following steps: step S1, collecting image information and realizing feature recovery of a dynamic fuzzy object through a fuzzy feature recovery module; step S2, guiding a model to focus on key features of an object based on a reinforced feature recognition mechanism, and recognizing potential dynamic objects; and step S3, jointly estimating the pose of a camera itself and judging the motion of an object to remove a dynamic object. The application can reconstruct and recover lost feature information from a fuzzy image, greatly improves the recognition accuracy of a system for a dynamic object, greatly improves the recognition accuracy of a dynamic object, and avoids misjudgment of static features.

The present application provides a blur processing method and device, wherein the blur processing method includes: obtaining an initial image; creating a reduced image of a corresponding blur level based on the initial image; determining a target image in the reduced image and determining a sampling offset of the target image in the pixel dimension based on preset blur information and pixel information of the initial image data; blurring the target image based on the sampling offset to generate a blurred image corresponding to the initial image.

The invention discloses a small sample hyperspectral image classification method and device based on fuzzy contrast graph convolution. The method comprises the following steps: determining membership information, fuzzy node initial feature vectors and similarity of each initial graph node corresponding to each fuzzy cluster based on a hyperspectral image; generating a fusion weight, a fusion feature and a target feature according to convolution features output by convolution of different layers of fuzzy images; performing linear transformation operation and probability distribution operation on the target features in sequence to obtain the probability that the corresponding initialized graph node belongs to each category; and performing argmax operation on the probability to obtain a category index corresponding to the hyperspectral image. The uncertainty of pixels is effectively processed through fuzzy learning, interference of neighborhood pixels is weakened, the uncertainty between the pixels is better processed, and meanwhile richer and more expressive feature representation is obtained. The method has higher classification effect and robustness, and plays an important role in multiple fields of mineral exploration, environment monitoring, forest management, precision agriculture and the like.

The invention relates to the technical field of image processing, computer vision and image blind deblurring, in particular to a self-supervised image blind deblurring method based on hierarchical Bayesian representation, which comprises the following steps: acquiring and processing a clear image, synthesizing a blurred image, and establishing a blurred image data set; constructing a hierarchical Bayesian prior model and a target function based on the motion degradation model in combination with the blurred image data set, and defining a random variable; the objective function is minimized, and posterior distribution of random variables in the layered Bayesian prior model is solved; a deep neural network framework is designed, and a loss function is established and trained in combination with posterior distribution of random variables to obtain an optimal model; outputting an image recovery result of the blurred image data set through the optimal model; according to the method, sparse adaptation optimization is realized on the basis of the hierarchical Bayesian prior model, the problems of insufficient generalization and poor robustness of a VB-based blind deblurring algorithm are solved, and the method can be applied to high-precision demand scenes such as image enhancement.

The invention relates to the technical field of data management, and discloses an enterprise sensitive data desensitization method and system based on a natural language, and the method comprises the steps: carrying out the classification processing of original enterprise data, obtaining classified enterprise data, carrying out the character feature extraction of enterprise text data, obtaining text feature characters, and carrying out the desensitization of enterprise sensitive data; evaluating character utility degrees corresponding to the text feature characters; marking desensitized feature characters in the text feature characters, and performing semantic fuzzification processing on the desensitized feature characters to obtain fuzzy feature characters; performing area blurring processing on the image desensitization area to obtain a blurred image area; desensitizing the enterprise numerical data to obtain target numerical data; and performing data updating processing on the classified enterprise data by using the target numerical data, the fuzzy image region and the fuzzy feature characters to obtain enterprise desensitization data corresponding to the original enterprise data. The main purpose of the invention is to solve the problem of low flexibility of enterprise sensitive data desensitization.

The invention relates to the technical field of image display, in particular to a method and device for generating a fuzzy layer of an auxiliary screen, a display screen and a vehicle. The screen is divided into a main screen and an auxiliary screen, the auxiliary screen is simulated through a virtual screen, and the auxiliary screen covers display content, located in an auxiliary screen display area, of the main screen; under the condition that a first image layer displayed on the auxiliary screen comprises a fuzzy image layer, the method comprises the steps that a second image layer which is located below the fuzzy image layer and has an intersection with the fuzzy image layer is determined, and the second image layer is an image layer on the main screen; under the condition that the second image layer meets a first condition, performing fuzzy processing based on rendering information of an application corresponding to the second image layer to generate a fuzzy image layer; and under the condition that the second image layer meets a second condition, obtaining a snapshot of the main screen, and performing fuzzy processing on the snapshot to generate the fuzzy image layer.

The present invention discloses an image-driven tick detection comb that integrates a deep learning model, which relates to the field of tick detection technology and includes an initial exposure time configuration module, an image data acquisition module, a feature extraction and fuzzy quantization module, a fuzziness prediction module, an image classification module, a normal image processing module, and a fuzzy image exposure adjustment module; the initial exposure time configuration module first determines the initial exposure time that is adapted to the current environment and detection requirements, and performs parameter configuration on the micro camera. The present invention uses deep learning and image processing technology to accurately predict image fuzziness, distinguish between normal and fuzzy images, and adopts adaptive exposure time control for fuzzy images to reduce motion blur, ensure image clarity, improve feature extraction accuracy, reduce the probability of misjudgment, detect and remove ticks in a timely manner, and reduce the risk of host infection with pathogens. Dynamic exposure adjustment enhances the robustness and reliability of the equipment in different environments, significantly improving detection accuracy.

The present invention relates to the technical field of image recognition, specifically a one-way blurred image registration method based on line features, comprising the following steps: confirming the clear direction of the one-way blurred image, extracting line features with an included angle less than 45° with the clear direction and matching them; confirming the linear equation coefficients of the matched line features and the lengths of the line features; constructing a matrix equation and introducing a line segment length weight coefficient; solving the coefficient X based on the least squares method in step S3 to obtain a homography matrix; obtaining the homography matrix H by solving the coefficient X according to the least squares method, and using the homography matrix for image registration. Since the one-way blurred image has higher clarity in the non-blurred direction, by extracting line features along the other clear direction, the extraction has higher accuracy. Compared with the method of calculating the intersection points of the straight lines where the two line segments are located after extracting the line segments and then using the intersection points to calculate the homography matrix, it has higher numerical stability and robustness.

The invention discloses a wafer annealing thermal management control system, and relates to the technical field of annealing thermal management. The method comprises the steps of calculating an error variable by acquiring temperature information of a target wafer in real time; determining a fuzzy image according to the error variable and a preset membership function; calculating a coefficient adjustment value according to the fuzzy image; setting the reference coefficient according to the coefficient adjustment value to obtain a target scale factor; and a PID control algorithm is adopted, and the heating power is calculated according to the target scale factor and the temperature difference value. Through combination of temperature acquisition, fuzzy control and self-adaptive PID adjustment, the problems of temperature response delay, overshoot oscillation, poor anti-interference capability and the like in traditional control are effectively solved. The system performs fuzzy reasoning based on the temperature difference and the change rate thereof, dynamically adjusts PID parameters, and optimizes an initial coefficient through a particle swarm and simulated annealing hybrid optimization algorithm, so that high-precision and high-robustness temperature control adjustment is realized, and the temperature control precision is remarkably improved.

The invention relates to the technical field of power grid safety protection, in particular to a collaborative false data injection attack detection method and device, electronic equipment and a medium. The method comprises the following steps: acquiring a current node feature and a current edge feature of a power system; based on the current node features and the current edge features, a first target feature matrix is generated by using a preset fuzzy graph convolutional network model, and the preset fuzzy graph convolutional network model is obtained by integrating a fuzzy learning module and a graph convolutional network; and inputting the first target feature matrix into a preset multi-agent detection module to obtain a current attack detection result of the power system. Therefore, the problems that limitation exists when an existing detection technology processes a complex non-linear relation, and accuracy and interpretability of a detection result are insufficient are solved, and therefore effectiveness and accuracy of collaborative false data injection attack detection are remarkably improved.

The application provides a fluorescence cell multilayer microscopic imaging method and system based on multi-feature fusion, and relates to the technical field of microscopic imaging. The method comprises the following steps: acquiring a multilayer Z-axis image sequence of a fluorescence cell sample at an X-Y scanning point; dividing an optimal focal plane image into blocks, and identifying the definition state through a convolutional neural network classification model; searching for a replacement block in a neighboring Z layer according to the classification result for a fuzzy block; and outputting a full field of view image after local registration and feathering fusion. The application solves the problems of unstable automatic focusing and insufficient single focal plane coverage under weak fluorescence conditions by fusing multi-feature definition evaluation and deep learning block classification, and realizes full field of view clear imaging of a fluorescence sample with uneven thickness.

The application relates to the technical field of vehicle speed identification methods, and discloses a fuzzy image frame analysis vehicle speed identification method based on field reproduction, wherein, according to the differences of a camera, a field environment and road conditions, a field reproduction positioning measurement method in the fuzzy image vehicle speed identification method is divided into a direct measurement method and an indirect measurement method. Through the use of the method, the limitations and deficiencies of target vehicle calibration distance selection in the fuzzy video image for vehicle driving speed identification are overcome, the problems of identification failure and inaccuracy are solved, most of the video image identification materials which cannot be identified by using the traditional method regain the use value, the problems caused by the defects of the identification materials are solved, the requirements for the identification materials are reduced, the problems of fruitless vehicle speed identification and inaccurate vehicle speed identification results are solved, the scientificity and accuracy of the identification opinions are improved, and strong evidence support is provided for the traffic police department for accident handling.

The embodiment of the invention provides a bokeh fuzzy processing method and device, equipment, a storage medium and a product. According to the technical scheme provided by the embodiment of the invention, the background mask graph is determined according to the transparency graph of the to-be-processed image, the index calculation is performed on the background mask graph and the color value of each pixel point in the to-be-processed image to obtain the light source spread function, and the convolution mask is determined according to the light source spread function and the background mask graph. And performing mask convolution processing on the to-be-processed image according to the convolution mask and a preset convolution kernel to obtain a background blurred image, and performing fusion processing on the to-be-processed image and the background blurred image according to the transparency image to obtain a target blurred image. The unnatural fusion at the foreground and background junction is effectively reduced, the color leakage of pixel blurring at the foreground and background junction is reduced, the natural degree of the edge during the foreground and background fusion is improved, and the blurring effect of the scattered scene is effectively improved.

The invention provides a denoising and non-uniform blurring-removing image enhancement method based on a multi-scale pyramid, and the method comprises the steps: firstly constructing a multi-scale Gaussian pyramid image, determining a blurring layer and a mask through fuzzy segmentation in order to improve the accuracy, carrying out the layered estimation of a motion blurring layer, and determining a blurring matrix of each image through a non-blind method, the method comprises the following steps: firstly, removing low-frequency noise by using a Gaussian low-pass filter, removing high-frequency noise by using wavelet transform, then performing deconvolution deblurring in a Gaussian pyramid, fusing two processing results through weighted multi-scale, and finally, performing inverse operation through a multi-scale pyramid to synthesize a clear picture. According to the method, the fuzzy kernel is solved in a layering mode, the Gaussian pyramid is used for denoising and deblurring, meanwhile, the high-frequency noise problem which is prone to being ignored in a traditional algorithm is considered, adaptability to pictures of different sizes is enhanced through the pyramid, uneven blurring caused by camera movement is removed, and the PSNR is increased by about 2.74 d compared with the PSNR in a traditional method.

This invention proposes a dynamic scene deblurring method and system based on 4D Gaussian representation and pseudo-ground value supervision. The method includes: obtaining a sharp image and a rendered depth map from a 4D Gaussian representation; predicting the motion velocity of dynamic pixels using a velocity MLP network based on the rendered depth map; constructing a fuzzy weight network and predicting the contribution weight of each sampling point on the sampling trajectory; fusing the pseudo-sharp image with the original blurred input using a dynamic region mask to obtain a hybrid pseudo-ground value image; synthesizing a physically blurred image through weighted integration; and constructing a dual-domain reconstruction loss and geometry-motion regularization constraints by combining hybrid pseudo-ground value supervision and gradient decoupling strategies to jointly optimize the scene representation, motion parameters, and fuzzy model, thereby obtaining a sharp dynamic scene representation. This invention achieves explicit modeling of the physical blurring process through 4D Gaussian representation and a fuzzy weight network, improving the realism and view consistency of motion blur synthesis.

The present invention relates to a network and method for image deblurring, and particularly to an image blind deblurring network and method based on fuzzy kernel prior learning. It solves the technical problem that the prior art regards image blind deblurring as an image regression task, without considering the differences in different data sets and the changes in local image fuzzy kernels, resulting in poor generalization performance. The network of the present invention includes a fuzzy kernel estimation network and a deblurring network. The fuzzy kernel estimation network is used to estimate the fuzzy kernel image of the blurred image, and it includes a feature extraction network, an uncertainty learning module, and a normalizing flow model. The deblurring network is used to obtain a clear image through the blurred image and the estimated fuzzy kernel image, and it includes a downsampling unit, an intermediate unit, and an upsampling unit connected in sequence. The downsampling unit includes a plurality of downsampling layers connected in sequence; the intermediate unit includes a plurality of basic modules connected in sequence; the upsampling unit includes a plurality of upsampling layers connected in sequence.

The application discloses a kind of fuzzy type self-adapting image restoration methods, belong to image processing and computer vision field, comprising: frame extraction processing is carried out to input video stream, and the image frame extracted is preprocessed;Based on preprocessed image, through multiple feature fuzzy evaluation mechanism, integrated fuzzy score is calculated;According to integrated fuzzy score, whether image frame is fuzzy frame is judged;If image frame is fuzzy frame, then according to multiple evaluation scores in multiple feature fuzzy evaluation mechanism, the fuzzy type of image frame is discriminated by preset rule;From multiple preset repair strategies, a corresponding repair strategy is selected to repair image frame, and a repaired image is generated;The quality of the repaired image is evaluated, and if the evaluation result does not meet the preset condition, the repair strategy is adjusted and re-repaired;The repaired image that meets the preset condition is re-inserted into the input video stream, and the repaired video is output.The application realizes the efficient identification and accurate repair of fuzzy image frame.