Enhanced Partial Distortion Sorting Fast Motion Estimation Algorithm for Low-Power Applications
Yang Song, Takeshi Ikenaga, Satoshi Goto, Zhenyu Liu · 2006
This paper presents an enhanced partial distortion sorting (EPDS) fast motion estimation (ME) algorithm which is based on the previous partial distortion sorting (PDS) algorithm. Different to PDS which has fixed parameters, in the EPDS algorithm, these parameters are adaptively adjusted to further reduce the computation complexity. Experiments show that the proposed EPDS algorithm has almost the same image quality as PDS but with 30.72% computation reduction. The PDS and EPDS are both realized in 1D systolic array, under the worst working conditions (1.62V, 125degC) and 166MHz clock frequency, compared with PDS algorithm, the proposed EPDS algorithm can reduce 15.75% power consumption with 0.75K gates hardware increase