12 results about "Residual frame" patented technology

The invention discloses a video compression intelligent preprocessing method, system and device based on frequency domain perception optimization and a medium. The method comprises the following steps: constructing a preprocessing model; preprocessing the input coding frame and the reference frame; enabling the preprocessed coding frame and reference frame to pass through a virtual encoder to obtain a frequency domain coefficient of a residual frame, and calculating a code rate of the coding frame; enabling the frequency domain coefficient of the residual frame to pass through a virtual decoder to obtain a decoded frame; training a preprocessing model; testing the effect; the system, the equipment and the medium are used for implementing the method. According to the method, a frequency domain sensing module is introduced into a virtual encoder in a training stage, modeling is carried out on a frequency domain information context, and statistical redundancy between transformation units is eliminated; an optical flow collaboration module is introduced into a virtual encoder, local distortion of a prediction frame is repaired in a feature domain by using high-level semantic information, and the difference between training and actual deployment is effectively reduced; through the optimization, the preprocessing model can better learn the rate distortion characteristic of the standard encoder, and the quality of the preprocessed video is remarkably improved.

Disclosed in the present invention are a method and system for target detection in a night scene, and a storage medium, which are applied to the technical field of computer vision, and are capable of effectively improving the accuracy of living-body target detection in night scenes, and improving the accuracy of positioning identified targets. The method comprises: performing preset decoding on a video to be subjected to detection, so as to obtain video frame images; performing grayscale conversion on the video frame images, and storing same in a grayscale frame array; when it is determined that the length of the grayscale frame array is greater than a first length threshold, acquiring grayscale frame image pairs from the grayscale frame array to perform differentiation operations, so as to obtain residual frame images; when pixel values of the residual frame images meet a preset pixel condition, storing the residual frame images in a second image array, so as to obtain a residual frame array; when it is determined that the length of the residual frame array is greater than a second length threshold, performing analysis on the basis of pixel points of the residual frame images in the residual frame array, so as to obtain target pixel coordinate points; and performing preset clustering processing on the basis of the target pixel coordinate points, so as to obtain a target identification result.

The invention relates to the technical field of computer vision and video processing, and provides a residual frame video processing convolution method based on dynamic gating, which comprises the following steps of: carrying out gray conversion and Gaussian filtering processing on a current frame image and a previous frame image; calculating an inter-frame difference image and carrying out binarization processing to obtain a motion area mask; performing morphological expansion operation on the mask image; extracting the contour of the motion area and calculating a minimum bounding rectangle to obtain a bounding box; carrying out merging processing on the overlapped bounding boxes; updating the synthesized background frame based on the combined bounding box, and copying the pixels of the motion area of the current frame to the corresponding position of the background frame; and outputting the optimized bounding box set and the updated synthesized background frame. According to the method, the calculation efficiency and the resource utilization rate of video processing are improved, and meanwhile, the timing sequence continuity and the space consistency of a processing result are ensured through a strategy of keeping the static region unchanged and only updating the motion region.

The present disclosure relates to methods, apparatus, systems, and non-transitory computer-readable storage media for video coding using super-resolution restoration with residual frame coding. According to some examples, a computer-implemented method includes receiving a coded frame of a video; performing a video coding on the coded frame of the video to generate a resultant for the coded frame at a second lower resolution than a first resolution; upsampling the resultant in at least a vertical direction to a higher resolution than the second lower resolution to generate an upsampled resultant; generating a decoded frame based on at least the upsampled resultant; and transmitting the decoded frame to a frame buffer or to a display device.

The invention discloses a video human motion recognition method based on a hybrid pulse neural network and an attention mechanism, and the method comprises the steps: carrying out the data preprocessing, dividing a video stream into a key frame and a residual frame, inputting the residual frame into a TA-LIF to generate a pulse sequence, extracting a time feature through an SNN network, carrying out the Relu function representation of the key frame, and carrying out the recognition of the human motion of a video. The method comprises the following steps: inputting ANN network branches to extract spatial features, carrying out attention mechanism fusion on time features extracted by an SNN network and the spatial features extracted by the ANN network to form fusion features, and carrying out action classification and optimization according to the fusion features, so that through the network design, the defects of high energy consumption of the ANN network and insufficient recognition accuracy of the SNN network are overcome, and the recognition accuracy of the SNN network is improved. The model calculation complexity is reduced, and the recognition accuracy is improved.

This disclosure provides methods, devices, and systems for video coding. The present implementations more specifically relate to hybrid coding techniques that combine aspects of inter-frame coding with aspects of intra-frame coding. In some aspects, a video encoder may perform inter-frame coding in a weighted manner so that the coded video frames (also referred to as “residual frames”) may include contributions from the current video frame to be transmitted over a communication channel and also may include contributions from the previous video frame transmitted over the communication channel. More specifically, any pixel value (r(n)) in the residual frame can be expressed as a weighted combination of a respective pixel value (x(n)) in the current video frame and a co-located pixel value (x(n−1)) in the previous video frame, where r(n)=x(n)−α·x(n−1) and where 0≤α≤1 is a scaling factor representing the degree of contribution by the previous video frame.

The invention discloses a design method for improving the transmission stability of a CoaXPress acquisition system, and relates to the technical field of machine vision. The invention provides a low-resource, reversible and self-adaptive data randomization mapping scheme aiming at the problems of signal baseline drift and error codes caused by the combination of a Nyquist code pattern and a random code pattern under 8B / 10B coding. The method comprises the following steps: when an acquisition card is idle, sending a detection feature code pattern DETECTWORD at an interval of 10ms; starting a mapping function after the camera detects the abnormal code pattern, performing byte-by-byte XOR processing on the content of the data packet containing the abnormal code pattern by adopting the mapping function, only processing the content, not moving the tail of the packet header, and changing the last K27.7 of the packet header into K29.7; and after the acquisition card identifies the packet header which is K29.7 finally, performing same function inverse XOR on the content to recover the original data and repair the packet header. According to the method, the bit error rate is remarkably reduced, the residual frame loss phenomenon is eliminated, and the system transmission stability is greatly improved on the premise that the hardware cost is not increased.

A method and an apparatus are disclosed for video coding using an improved in-loop filter. The video coding method and the apparatus generate a residual frame from a reconstructed frame using a deep learning model. The video coding method and the apparatus improve performance of an in-loop filter by approximating an original residual frame by applying the generated residual frame to a linear model.

A method and an apparatus are disclosed for video coding using an improved in-loop filter. The video coding method and the apparatus generate a residual frame from a reconstructed frame using a deep learning model. The video coding method and the apparatus improve performance of an in-loop filter by approximating an original residual frame by applying the generated residual frame to a linear model.

The invention provides a signal data processing method and device, equipment and a storage medium, and relates to the technical field of image processing, and the method comprises the steps: determining whether a time domain recovery flag bit is set in code stream data, determining the number of residual frames under the condition that the time domain recovery flag bit is set, and carrying out the frame compensation based on the number of the residual frames, and obtaining a first complementary frame segment. Coding and decoding based on variable up-sampling multiplying power are realized.

In various embodiments, a localized video encoding application encodes localized videos. In operation, the localized video encoding application computes a prediction frame based on a target frame of a localized video and at least a portion of a reference frame of a decoded original video. The localized video encoding application computes a residual frame based on the prediction frame and the target frame of the localized video. The localized video encoding application performs one or more encoding operations on the residual frame to generate a frame of an encoded localization layer. The frame of the encoded localization layer and at least one frame of an encoded original video are subsequently transmitted to another device for decoding.

In various embodiments, a localization video encoding application encodes a localization video. In operation, the localization video encoding application calculates a predicted frame based on a target frame of the localization video and at least a portion of a reference frame of a decoded original video. The localization video encoding application also calculates a residual frame based on the predicted frame and the target frame of the localization video. The localization video encoding application also performs one or more encoding operations on the residual frame to generate a frame of an encoded localization layer. Then, the frame of the encoded localization layer and at least one frame of the encoded original video are sent to another device for decoding.