This paper presents a novel approach to pixel decimation for motion estimat
ion in video coding. Early techniques of pixel decimation use regular pixel
patterns to evaluate matching criterion. Recent techniques use adaptive pi
xel patterns and have achieved better efficiency. However, these adaptive t
echniques require an initial division of a block into a set of uniform regi
ons and therefore are only locally-adaptive in essence. In this paper, we p
resent a globally-adaptive scheme for pixel decimation, in which no regions
are fixed at the beginning and pixels are selected only ii they have featu
res important to the determination of a match. The experiment results show
that when no more than 40 pixels are selected out of a 16x16 block, this ap
proach achieves a better search accuracy by 13-22% than the previous locall
y-adaptive methods which also use features.