Hybrid algorithm for block matching motion estimation technique
Saurabh P. Asare, Aniket V. Gokhale, Chetankumar Selukar, Shailesh D. Kamble · 2015
Motion estimation (ME) is one of the most time consuming part in video coding. But it is an effective method used for reducing temporal redundancies between two successive frames. It is also called as motion compensation technique (MCT). Many block-based fast motion estimation algorithms have been proposed till now. This algorithm efficiency depends on low-resolution coarse search and fine resolution inner search to recognize slow and fast movements of the object. This paper is suggesting a hybrid motion compensation technique (H-MCT) based on Efficient Three Step Search (ETSS) and Hexagonal Search (HS). This Hybrid algorithm effectively detects the slow and fast movements with less computation time. And the performance of this proposed scheme is evaluated in terms of two matching criteria, Sum of Absolute Difference (SAD) and Mean of Absolute Difference (MAD). Results show that proposed algorithm performs better than algorithms such as ETSS and HS using less no. of search points. All experimentation is carried out on standard database Akiyo.avi.