Comparative Study of Block Matching Techniques Used in Video Images Motions Estimation

摘要

The purpose of this work consists in a comparative study between block matching search algorithms, used in video images motion estimation. Exploiting the temporal correlation between two successive sequence frames we can reduce enormously the memory space needed for image registration. In this context, we have applied and compared different algorithms of search used in block matching technique. These algorithms include full search algorithm, three steps search algorithm, new three steps search algorithm, four steps algorithm, diamond search algorithm and minimum search algorithm. The performance of an algorithm is a compromise between the pick signal to noise ratio and the computing search time of the algorithm.