Video encoding inter-frame prediction method based on multiple linear regression

Abstract

The invention provides a video encoding inter-frame prediction method based on multiple linear regression. The method comprises the following flows: A, initializing an image at a moment t of a signal source, and using the image at a moment t-1as a reference image, and performing movement estimation on the image at the moment t at the moment t-1 to obtain an absolute error value; B, expanding the size of a reference pixel block at the moment t-1, wherein the expanded reference pixel block information is a left outer side pixel and an upper outer side pixel of the reference pixel block to obtain a multiple linear regression model; C. reconstructing the image at the moment t prior to the moment t to obtain a pixel value of a corresponding position in the reconstructed image at the moment t; and D, determining whether to update the reference image; if the absolute error value of the reconstructed pixel block and the corresponding signal source pixel block is smaller than the absolute error value in step A, updating the pixel of the reference image; and on the contrary, not updating the pixel of the reference image. By adoption of the video encoding inter-frame prediction method, the time domain correlation between the reference image and the current encoding image is effectively enhanced, then better residual information is obtained, and the encoding efficiency is obviously improved.

Description

technical field [0001] The invention relates to a video coding method, in particular to a video coding inter-frame prediction method based on multiple linear regression. Background technique [0002] Inter-frame prediction refers to using the correlation of the video time domain to predict the pixels of the current image using the pixels of the adjacent encoded image, so as to effectively remove the redundancy of the video time domain. At present, the inter-frame prediction part of the main video coding standards adopts the block-based motion compensation technology. The main principle is to find the best matching block in the previously coded image for each pixel block of the current image. This process is called motion compensation. Estimation (Motion Estimation, ME), where the image used for prediction is called a reference image (Reference Frame), and the difference between the current pixel block and the reference block is called a prediction residual (Prediction Residu...

AI Technical Summary

Problems solved by technology

[0003] In the encoding process, the texture complexity of the image content and the intensity of changes between images will affect the quality of the encoded image, and the method of using the block as the coding unit will lead to uneven pixels at the boundary of the image block. These factors will Affecting its own reference will make the predicted residual information unfavorable for coding

Images