High Performance Multicore Implementation of Motion Picture Estimation

Motion Estimation in video transmission helps to reduce the memory consumption with slight difference in the retrieved video signal with respect to the original video signal. Some of the applications of Motion Estimation are in Computer Vision, Image Sequence Analysis, Video Coding, and Target Tracking. In this paper, we have proposed multi-threading technique for calculation of motion vectors of multiple video frames in parallel to reduce the latency. The number of threads used for computational purpose, the more it reduces the time for execution. Using 12-cores implementation, the implementation results in increase of 34.60% compared to the sequential execution that uses no multi-threading technique and only uses a single thread computation that is the conventional method.