A New Block Motion Estimation Algorithm
Jun Ge and G. Mirchandani
ICASSP2002, Orlando, FL
(Paper (pdf) / Paper (ps))
A fast and efficient block-matching motion estimation algorithm is proposed in
this paper. By using the fact that the
motion characteristics are relatively stable in a GOP (group of pictures), the
proposed algorithm produces fast computation
and low bitrates while keeping the high PSNR. The motion estimation of the P-f
rame is performed using a modified exhaustive
search algorithm to guarantee the extracted motion information for the GOP is
reliable. Then the extracted motion
information is utilized to classify the macroblocks into compensable and uncom
pensable, and to adapt the search window size
for the motion estimation of the B-frames in that GOP. The performance and the
computation complexity are compared with
other block-matching algorithms.