Motion compensation structure in multi-mode video decoder

Abstract

The invention discloses a motion compensation structure in a multi-mode video decoder. The motion compensation structure comprises a frame storage for storing a reference image; a prefetching cachingunit for reading and caching reference pixel of the external frame storage; a coefficient page table for storing a filter coefficient of a video decoder suitable for various standards; an interpolation unit comprising FIR filters capable of configuring parameters, wherein the interpolation unit is used for performing horizontal direction 1 / 2 interpolation, vertical direction 1 / 2 interpolation, horizontal direction 1 / 4 interpolation and vertical direction 1 / 4 precision interpolation filtering; an output end for outputting brightness interpolation data and chromaticity interpolation data, wherein the prefetching caching unit is connected with the frame storage; and the interpolation unit is connected with the prefetching caching unit, the coefficient page table and the output end at the sametime. By adopting the FIR filter with configurable parameter, the system can take the interpolations of both of the brightness component and the chromaticity component in various video standards intoconsideration. The invention further discloses a motion compensation method based on the system.

Description

technical field [0001] The invention relates to the technical field of multimedia video, in particular to a motion compensation structure in a multi-mode video decoder, and a motion compensation system and method that meet the decoding requirements of high-definition video on the basis of low power consumption. Background technique [0002] Motion compensation technology improves the efficiency of video coding by eliminating temporal redundancy between adjacent frames. In the latest video coding standards, motion compensation introduces many new coding techniques, including variable block size, multiple reference frames, bidirectional prediction, unrestricted motion vectors, and quarter-precision interpolation, etc. All these new technologies can improve the efficiency of video coding, but at the same time greatly increase the computational requirements and bandwidth requirements of motion compensation. In the basic level of H.264 / AVC, motion compensation occupies 25% of th...

AI Technical Summary

Problems solved by technology

[0003] Previous high-efficiency motion compensation circuits for H.264 / AVC all adopted a motion compensation structure based on a 4x4 fixed block size, and proposed a complex data reuse strategy and scheduling strategy to overcome the motion compensation based on a 4x4 block size. The defects brought about by these new strategies lead to the need for more on-chip memory and complex control logic to support these new strategies. At the same time, different hardware interpolation circuits are used for the brightness component and the chrominance component; the previous structure can support H.264 at the same time / AVC and AVS luma interpolation structure, but does not support chroma interpolation, and does not support other standards; some structures mainly focus on the design and implementation of motion compensation subsystem in multi-standard decoders, including motion vector prediction, etc. Does not contain a detailed description of the motion compensation interpolation structure; someone proposed to use a new 4-level directed FIR filter to replace the original 6-level filter in the H.264 / AVC standard to solve the problem of the original interpolation algorithm. Regulatory issues and excessive bandwidth requirements, but it introduces a serious loss of image quality, and also fails to propose a solution suitable for multiple standards

Images