A Pattern-Based Artificial Bee Colony Algorithm for Motion Estimation in Video Compression Techniques

摘要

Block matching (BM) motion estimation plays an inevitable role in video coding applications. BM approaches are used for data compression. The compression is achieved by removing the temporal redundancy in the video sequences. In the BM process, each video frame is subdivided into macroblocks. Each macroblock in the current frame is compared with the previous frame. The main objective is to minimize sum absolute difference. In this work, some modifications have been performed on conventional artificial bee colony algorithm to improve the conventional BM systems. An initial pattern is used in the proposed algorithm to reduce the computational cost. The computational cost is represented in terms of search points and convergence time. Experimental results results show the improvement for the proposed approach over other block matching algorithms in terms of the performance measures.