Sub-pixel motion estimation method based on AVS (audio video standard)

Abstract

The invention relates to a sub-pixel motion estimation method based on AVS (audio video standard), and the method comprises the following steps: 1) performing the motion vector search on 1 / 2 pixel by utilizing a full-pixel searching result and confirming the position of the optimum 1 / 2 pixel point; 2) if the most matching 1 / 2 pixel point is a central full-pixel point, judging whether the matched function values of other candidate points are more than a threshold T, and if yes, no longer performing the 1 / 4 pixel motion vector search, otherwise performing a step 3); and step 3) performing the motion estimation forecasting on 1 / 4 pixel points around the optimum 1 / 2 pixel point, thereby confirming the optimum 1 / 4 pixel point. According to the method, the area of the 1 / 2 pixel point is forecasted according to a matching error value of the optimum full-pixel point and the full-pixel points around the optimum full-pixel point and the search is ended in advance through the threshold value judgment, thereby ensuring the accuracy of fraction pixel search, effectively reducing the number of search points and simultaneously maintaining the coding efficiency as much as possible.

Description

technical field [0001] The invention belongs to the technical field of audio and video coding and decoding, and in particular relates to a sub-pixel motion estimation method based on the digital audio and video coding standard (Audio Video coding Standard, AVS). Background technique [0002] In AVS, H.264 and other standards, the computational complexity of inter-frame motion estimation and intra-frame prediction is relatively high, especially the motion estimation part, which usually accounts for more than 50% of the total encoding time. Motion estimation consists of two parts: integer-pixel motion estimation and sub-pixel motion estimation. With the development of fast algorithms for integer-pixel motion estimation, the number of search points for integer pixels is greatly reduced. Generally, it can be reduced to less than 10 points, resulting in sub-pixel motion estimation. The amount of calculation accounts for an increasing proportion of the calculation amount of the en...

AI Technical Summary

Problems solved by technology

[0004] Although these methods have greatly improved compared with the full search algorithm, the algorithms are relatively complex and are not conducive to the implementation on the hardware platform. Therefore, reducing the complexity of the sub-pixel motion estimation algorithm and shortening the motion estimation time is a top priority.

Images