76 results about "Video restoration" patented technology

The invention provides an electronic image stabilization method based on characteristic coupling. The method comprises the following steps: (1) motion estimation: estimating motion parameters of a present frame and a previous frame to obtain a global motion vector and a local motion vector; (2) motion filtering: employing a Kalman filter to carry out filtering on the global motion vector, and obtaining a willful scan vector, wherein difference between the global motion vector and the willful scan vector is a vibration parameter needs to compensate; (3) motion compensation: utilizing a compensation parameter obtained by filtering carrying out frame by frame compensation; (4) video restoration: by utilizing space-time redundant information of video, carrying out restoration on the local motion vector with a motion restoration method and guiding pixel point filling. In order to solve motion carrier vibration influence on video shooting effect, the invention provides the electronic image stabilization method based on characteristic coupling. The method has the advantages of high accuracy, good anti-noise performance, strong tracking performance, good boundary consistency and the like.

The invention provides a video super-resolution processing method and device, and the method comprises the steps: obtaining each frame of image of a video, and inputting the image into a convolutionalneural network; Sequentially performing feature extraction, feature dimension reduction, nonlinear mapping and high-dimensional space mapping on the image through the convolutional neural network toobtain super-resolution features; Performing reconstruction according to the features obtained by feature extraction and the super-resolution features to obtain a super-resolution image; And finally,performing coding to form a super-resolution video code stream. The super-resolution processing of the video is realized through the convolutional neural network; Dimension reduction through features,nonlinear mapping and high-dimensional space mapping are carried out; the calculation complexity is reduced, the time complexity is reduced, the network learning difficulty is reduced and the complextexture of the output image is reserved by adopting the jump connection, so that the high reconstruction quality is realized while the real-time performance required by video processing is ensured, and the method has a very wide application prospect in the fields of video real-time transmission and compression, video restoration and the like.

The invention discloses a device eliminating interference of a television on a speech recognition device, which comprises an audio receiver connected with the audio output end of the television, an HDMI audio and video separator connected with the HDMI audio and video output end of a set top box of the television, a video output unit, a first audio output unit, an audio selector, an echo elimination unit, a speech enhancement and noise reduction unit, an audio acquisition unit, a second audio output unit, and a playback module, wherein the echo elimination unit is connected with the audio selector, the audio acquisition unit and the speech enhancement and noise reduction unit; and the audio selector is connected with the HDMI audio and video separator and the audio receiver. Technologies such as television audio signal introduction, HDMI audio and video separation, television video restoration, television audio restoration, audio selection, echo elimination, speech enhancement and noise reduction and audio acquisition are combined together, a new technical scheme in which the television and an intelligent speech device can be used at the same time is formed, and the problem that the television and the intelligent speech device can be used at the same time can be solved.

The invention relates to a video restoration method and apparatus, an electronic device and a storage medium. The method comprises the steps of obtaining a first forward optical flow and a first reverse optical flow between adjacent images in multiple continuous frames of images; performing optical flow optimization processing on the first forward optical flow and the first reverse optical flow respectively to obtain a second forward optical flow corresponding to each first forward optical flow and a second reverse optical flow corresponding to each first reverse optical flow; carrying out forward conduction optimization on the continuous multi-frame images by utilizing the second forward optical streams, and carrying out reverse conduction optimization on the continuous multi-frame imagesby utilizing the second reverse optical streams until all the images in the optimized continuous multi-frame images meet repair requirements; and according to the optimized images obtained through the forward conduction optimization and the reverse conduction optimization, obtaining the continuous restored image of the multi-frame image. According to the embodiment of the invention, the consistency of the continuous frame repair and the real-time performance of repair can be improved.

The invention discloses a video encoding method, a video restoration method, a video playing system and related devices such as a server and a playing device, and relates to the technical field of computers. The video encoding method disclosed by the invention comprises the following steps: segmenting a to-be-encoded video into corresponding sub-videos according to the resolution of a plurality ofencoding devices, wherein the resolution of the sub-videos is lower than that of the to-be-encoded video; and separately inputting the sub-videos in the plurality of corresponding encoding devices for encoding. According to the video encoding method disclosed by the invention, the high-resolution to-be-encoded video is cut into a plurality of low-resolution sub-videos according to the resolutionof the plurality of encoding devices, and the sub-videos are encoded via the plurality of encoding devices, the encoding of the high-resolution video is achieved by superposing a plurality of low-resolution encoding devices, so that the improvement of the resolution of the video can be adapted by using the existing devices without upgrading or replacing the encoding devices, thereby reducing the cost and saving resources.

The embodiment of the invention provides a video restoration method and device and electronic equipment, and belongs to the technical field of data processing, and the method comprises the steps: responding to specific annotation information related to video restoration, and obtaining a target area set for an initial frame of a video file; in all video frames of the video file, based on an object edge constructed by a gradient value of the video frame, carrying out tracking segmentation processing on the target object by utilizing a segmentation network comprising a foreground channel and a contour channel which are distributed in parallel; executing video restoration operation on all video frames in the video file based on the result of the tracking segmentation processing; and obtaining pixel points contained in the video frame formed after the restoration operation is executed, and carrying out smoothing processing on the restored video frame according to a preset smoothing processing curve to form a restored video. Through the scheme of the invention, the video restoration efficiency is improved.

The embodiment of the invention provides a video restoration method and device, equipment and a storage medium, and relates to the technical field of artificial intelligence, and the method specifically comprises the steps: employing a neural network model to extract the image features of a to-be-restored video frame, and predicting the image compensation parameter value of the to-be-restored video frame based on the image features. And dividing the to-be-restored video frame into a plurality of to-be-compensated regions according to the distribution characteristics of the video coding parameters corresponding to the to-be-restored video frame. And determining a region compensation parameter value of each region to be compensated according to the type of each region to be compensated, a preset image compensation strategy and the image compensation parameter value. And performing image compensation on each to-be-compensated region based on the region compensation parameter value of eachto-be-compensated region to obtain a restored video frame. Dividing the to-be-restored video frame into the plurality of to-be-compensated regions, determining the region compensation parameter valueof each to-be-compensated region based on the predicted image compensation parameter value, and carrying out targeted image compensation, so that the video restoration effect is improved.

The invention discloses a real-time video image automatic de-duplication splicing method and system, and the method comprises the steps: deploying a plurality of cameras in a target scene, dividing the target scene into a plurality of key monitoring regions, and each camera corresponding to one key monitoring region; wherein marks are arranged at four corners of each key monitoring area, an overlapping area existing between two adjacent key monitoring areas, and two marks being shared by the two key monitoring areas in the overlapping area; collecting multi-channel videos of the plurality of cameras, and storing the multi-channel videos in a centralized manner; identifying a mark of each path of video, correcting each path of video according to a mark form and pixel coordinates, and extracting an image of a key monitoring region; and analyzing and de-duplicating the multiple paths of videos according to the marks in sequence, splicing and restoring the key monitoring regions of the multiple paths of videos into one path of video, and restoring the target scene. According to the automatic de-duplication splicing of the invention, ideal spliced images can be obtained for a pluralityof static cameras or a plurality of moving cameras.

The invention provides a video restoration method, a video restoration device and a server. The method comprises the steps of determining a mask of each frame of image in a to-be-restored video; wherein the to-be-restored video comprises a to-be-removed object; covering the to-be-restored region in each corresponding frame of image in the to-be-restored video according to the mask of each frame ofimage to obtain a first video; inputting masks of each frame of image in the first video and the to-be-restored video into a first neural network model to obtain a second video; wherein the to-be-restored region in each frame of image in the second video is filled with background information of other regions except the to-be-restored region; and synthesizing the to-be-restored region in each frame of image in the second video with other regions except the to-be-restored region in each frame of image corresponding to the to-be-restored video to obtain a restored video. According to the method,the video restoration efficiency is improved.

The invention discloses a video compression method based on deep learning, and belongs to the field of video compression and deep learning. The technical problem to be solved by the invention is how to realize video compression and storage, smaller storage is realized, and the better video recovery effect is obtained. The technical scheme is adopted as follows: the method comprises the steps thatmotion estimation and motion compensation calculation are conducted through a Spynet motion estimation network composed of an optical flow network, a motion vector coding network and a motion vectordecoding network, and the better motion estimation and motion compensation effect is achieved; a residual network is used again, the residual network comprises two Resblock modules, and network training is carried out under the condition of a deeper level; and coding is completed through arithmetic entropy coding operation, the codes are stored as a Pick file, compression and storage of the videoare achieved, the smaller storage is achieved, and a better video restoration effect is obtained.

The invention discloses an ESRGAN-based video super-resolution restoration technology, device and equipment, and a storage medium. The method comprises the steps of obtaining a to-be-restored video; extracting a key frame in the to-be-restored video, and restoring an image of the key frame; grouping video frames in the to-be-restored video to generate frame groups, wherein the frame group comprises a key frame; sampling the repaired key frame in the frame group to generate a sample block; and repairing the to-be-repaired frame in the frame group through the sample block. According to the method, the data processing amount in the video restoration process is relatively reduced, so that the overall efficiency of video restoration is improved. In addition, the invention further provides a video restoration device and equipment and a storage medium, and the beneficial effects are the same as described above.

A low-latency resolution adaptive video fiber transmission codec device comprises a video input end and a video output end, wherein the video input end comprises USB equipment A, an FPGA A, an interface module A, a USB interface chip A, and a photoelectric module A; the video output end comprises a photoelectric module B, an FPGA B, an interface module B, a USB interface chip B, and USB equipmentB; the FPGA A programs to generate a video processing module, a data frame closing module A, a data de-framing module A, and an EDID processing module A; the FPGA B programs to generate a data de-framing module B, a data frame closing module B, a clock module B, a video restoration module, and an EDID processing module B; the video input end obtains video data and control timing by real-time synchronous waveform sampling of the FPGA A, and re-encodes the video data and control timing to enable synchronous transmission of video encoding and timing encoding; and the video output end performs video synchronous restoration on the received video encoding and timing encoding through the FPGA B to achieve the purpose of low delay and resolution adaptation.

The invention provides a video restoration method, and belongs to the field of communication and information systems. The method mainly comprises the following steps of: extracting motion informationin an image frame sequence based on a mask corresponding to an original video containing a missing area; arranging a cubic sliding block for traversing and sliding in three dimension directions of theframe width, the frame height and the frame number of the original video; when the cubic sliding block slides to a mark region, randomly initializing a larger cubic sliding space; comparing the motion information with the optical flow characteristic information of the edge of the mask region, finding the most matched sample block in the sliding space range, filling and reconstructing the pixel characteristics of the mask region, continuing to traverse until all mask regions in the original video are restored, inputting the restored mask regions into a forward-backward neural network for optimization, and outputting a restored complete video. According to the method, a high-quality video restoration effect is achieved at the minimum calculation cost.

The invention discloses a scene video restoration device based on front-end object detection, comprising an object detection module, a frame sampling adjustment module and an image synthesis module. The object detection module is used for respectively detecting the object of each frame of an input video stream and sending the detected object data to the image synthesis module. The frame sampling adjusting module is used for extracting the background image from the input video stream and sending the background image and the attribute information corresponding to the background image to the image synthesizing module, and adjusting the rule of extracting background image; The image synthesizing module is used for synthesizing the target image and the background image to obtain the restored video stream. This design can ensure the accurate restoration of scene video, greatly reduce the bandwidth required to transmit images, and effectively save data storage space. In particular, this design can accurately identify and extract specific objects in the scene, in order to achieve accurate reproduction of scene details.

The invention relates to a video restoration method, a video restoration device, computer equipment and a storage medium, and belongs to the technical field of video processing. The method comprises the following steps: acquiring a to-be-restored video frame; determining the image quality damage degree of the to-be-restored video frame; if the image quality damage degree of the to-be-restored video frame is greater than or equal to a preset damage degree threshold, performing first restoration processing on the to-be-restored video frame through at least one restoration algorithm according toa serious damage restoration sequence; and if the image quality damage degree of the to-be-restored video frame is smaller than a preset damage degree threshold, performing second restoration processing on the to-be-restored video frame through at least one restoration algorithm according to the light damage restoration sequence. By adopting the method and the device, after the to-be-restored video frames are distinguished, the to-be-restored video frames can be restored through at least one restoration algorithm according to different restoration sequences. The restoration processing is relatively targeted, and the image quality of the obtained restored video frames is relatively high.

The invention discloses a fast image smog removing system based on a depth convolution neural network, which constructs an efficient single image smog removing depth convolution neural network model and a smog image training set, trains the network proposed by the invention, and finally reconstructs a clear smog-free image. By constructing deep convolution neural network model and atmospheric scattering model and training them in depth, the smog removal effect can be obtained quickly and the real-time application can be realized. The system of the present invention can be easily extended to other image or video processing fields, such as image or video deblurring, image or video restoration, and the like.

The invention discloses a video decompression method and system based on Hog features and an lgb classifier, and relates to the technical field of video decompression. The method comprises the steps of converting a monitoring video of an unmanned convenience store into a plurality of frames of images; selecting a first frame image of the original monitoring video, performing pedestrian detection by using the Hog features and the lgb classifier, and removing the pedestrian positions to form a background picture without a moving object; carrying out the pedestrian detection by using the Hog features and the lgb classifier to find out the pedestrian positions, and cutting the pedestrian positions as the pedestrian pictures to be stored and named; comparing each frame of image contained in theoriginal video with the background image to obtain a difference value, storing the residual data in a linked list, and compressing the residual data; taking out the background picture, and decoding and restoring the background of each frame of image according to the residual data in the linked list; taking out the pedestrian images in the linked list to cover each frame of background image, and decoding and restoring all frames of images; and splicing all the covered frame images into a video according to the bit rate and the code rate of the original video to complete the video restoration.

The invention relates to a video restoration method, apparatus and device, and a storage medium. Belonging to the technical field of multimedia, the embodiment of the invention provides a method capable of repairing a video. A video abstract of a video is subjected to resolution amplification processing; according to the method, the high-resolution target video abstract is obtained by extracting the high-resolution image of the video, the high-resolution image similar to each frame of image of the video is found from the target video abstract, the corresponding image in the video is restored through each frame of high-resolution image, and then the whole video can be restored by restoring each frame of image of the video. Through the method, information contained in the high-resolution image is fully utilized, and damaged details in the original image of the video are supplemented, so that the blurred video is converted into a clear video, and the quality of the video is effectively improved.

The invention relates to a method and a system for inserting image frames between videos and an electronic device. The method comprises the following steps: respectively selecting m frames of featuremaps containing pedestrians before video loss time and after video restoration time, and respectively collecting a set number of human body posture points from each feature map; inputting all the human body posture points into an Alex Net network, wherein the Alex Net network predicts pedestrian postures of the to-be-restored image by using a method of combining cubic polynomial fitting and cubicspline interpolation; inputting the human body posture points corresponding to the feature map before the video loss time into an LSTM network to obtain a pedestrian posture prediction result of the to-be-restored image; and obtaining a to-be-restored image according to pedestrian posture prediction results of the Alex Net network and the LSTM network, calculating an insertion position of the to-be-restored image in the video, and inserting the to-be-restored image into a corresponding position in the video. The precision of an existing algorithm is improved.