Adaptive Hexa-Diamond Search (AHDS) Algorithm for Fast Block Matching Motion Estimation

In this paper, a simple fast block matching algorithm based on Adaptive Hexa-Diamond Search (AHDS) is proposed to estimate motion vector parameter. This search consists of two sequential search stages: 1) initial search 2) refined local search. For initial search stage, hexagonal pattern is proposed to reduce the computational complexity, in which the least error is determined. The point that has the least error becomes the origin for subsequent refined local search steps, and the search pattern is changed to Small Diamond Search Pattern (SDSP) until the final motion vector (MV) is found. Based on the experimental results, AHDS needs only 4% of the total computations compared to Full Search (FS) algorithm and is very close to Adaptive Rood Pattern Search (ARPS). The result shows that time saved is 93.24% compared to FS algorithm. AHDS gives performance close to ARPS in terms of computational complexity, processing time and similar results as it is in other existing algorithms in terms of image quality.