Successive accumulating partial distortion search algorithm for motion estimation
Chia-Hung Yeh, Ming Wu, Shiunn Jang Chern · Tamkang University Institutional Repository (TKUIR) · 2010
This work presents a modified partial distortion search algorithm that considers the property of successive accumulation for motion estimation. In the normalized partial distortion search algorithm, the partial distortion is calculated by regularly selecting a certain number of pixels from every sub-macroblock for early termination. The proposed method calculates the partial distortion by successively accumulating the distortion of each partial sub-macroblock. In addition to increasing the probability of the rejection of invalid motion vectors without additional computations, the proposed method has a markedly lower computational complexity than conventional algorithms, but slightly poorer objective quality of motion compensated images.