Fast Implementation of Block Motion Estimation Algorithms in Video Encoders

This project will involve analysing some of the newer fast algorithms in comparison with an exhaustive full search algorithm for motion estimation in video encoders. It will investigate the issues of computational complexity and efficiency of each fast algorithm in relation to the quality of the decoded video frames. Motion estimation and compensation forms a significant contributor in the video compression process. Motion estimation involves prediction of a video frame from other previously decoded frames. Block motion estimation algorithms adopt various search criteria and techniques for constructing optimal predictions of the frames. Many different fast block matching algorithms (BMA) have been proposed over time. Combining each algorithm with the streaming SIMD extensions (SSE) is expected to increase processing speeds. Index: Motion estimation, Block motion estimation algorithms, Fast algorithms, Streaming SIMD extensions

• Type: Journal
• Year: 2011
• Where: CCP
• Authors: Naga Rohini Koduri, Mqhele E. Dlodlo, Gerhard de Jager, Keith L. Ferguson
• Keywords: CCP 2011, Motion Estimation, Optimal Predictions, Streaming SIMD Extensions