A Video Coding Method Combined with Video Stabilization

Abstract

The invention relates to a video encoding method combined with video stabilization, comprising the following steps: performing stabilization processing on the original video, and extracting video stabilization information; calculating a transformation matrix between stabilized video frames; calculating each macroblock in each stabilized frame Corresponding motion vector; use above-mentioned motion vector as the starting point of corresponding macroblock motion search, search each macroblock, obtain the optimal motion vector of each macroblock; obtain the optimal motion vector of each sub-macroblock; select matching The macroblock segmentation method with the least cost performs motion compensation, quantization and entropy coding to obtain the final stable video compression stream. The invention obtains more accurate motion vectors with less calculation amount, and greatly improves the coding speed while ensuring the video coding quality.

Description

technical field [0001] The invention relates to a video encoding method, in particular to a video encoding method combined with video stabilization, and belongs to the field of video processing. Background technique [0002] Video coding technology is a video compression technology, which mainly achieves the purpose of compressing video by removing spatial redundancy and temporal redundancy existing in video. Now the more common codec standards include ITU's H.264 standard, H.265 standard, Motion Picture Experts Group's MPEG standard, Microsoft's WMV standard, and so on. Among them, the use of H.264 is the most common. In the H.264 video coding standard, each frame of video is divided into macroblocks of a specific (in pixels) size (16x16, 16x8, 8x16, 8x8, 8x4, 4x8, and 4x4). Each macroblock is encoded either intra-frame or inter-frame. Both intra-frame coding and inter-frame coding require a previously coded video frame as a reference frame. For intra-frame coding, the ...

AI Technical Summary

Problems solved by technology

Although this type of method combines video coding and video stabilization, the existing methods all obtain the motion vector corresponding to the macroblock through the median prediction of the adjacent coded macroblocks in this frame, without multiplexing the video stabilization process. The generated transformation matrix information makes the accuracy of the generated motion vector low, which reduces the efficiency of the motion estimation process in video coding; and in order to ensure that each macroblock has at least one corresponding motion vector, so for each macroblock Each block requires at least one motion search process

Images