Real-time Motion Estimation for 1080p Videos on Graphics Processing Units with Shared Memory Optimization

Shang-Te Yang,  Tsung-Kai Lin,  Shao-Yi Chien
National Taiwan University, Taiwan


Abstract

Motion Estimation plays a fundamental role in many applications. A shared-memory-optimized implementation of motion estimation is studied in this work, and experimental results shows that at most 38 times speedup can be achieved. Moreover, based on the optimization technique, a fast algorithm for real-time motion estimation for 1080P videos is also proposed in this paper.