應用於H.264中，區塊匹配的快速演算法

Abstract

H.264/AVC是目前最新的影像壓縮標準，其中的動態評估(Motion Estimation)會占據70%左右的編碼時間，因此這部分是一個很重要的改良課題。這篇論文將提出一個新的演算法，它將結合DAS(Directional Asymmetric Search)以及SPS(Search Pattern Switching)兩種方式的特性，並改良搜尋Pattern的部分，使新的演算法可以比DAS以及SPS還快速，並保持著不錯的影像品質。

H.264/AVC is the latest video coding standard at this moment. Motion estimation, which is a part of this standard, occupies about 70% of the total encoding time. Therefore, it is an important section that needs to be improved. In this thesis, we propose a new fast block-matching algorithm, which combines the Directional Asymmetric Search algorithm (DAS) and the Search Pattern Switch algorithm (SPS) and improves the motion estimation section with adding search patterns in Directional Asymmetric Search. Our algorithm can cut the total encoding time compare with DAS and SPS, but still keep the similar PSNR and bitrates.