Our starting point is gradient indexing, the characterization of texture by
a feature vector that comprises a histogram derived from the image gradien
t field. We investigate the use of gradient indexing for texture recognitio
n and image retrieval. We find that gradient indexing is a robust measure w
ith respect to the number of bins and to the choice of the gradient operato
r. We also find that the gradient direction and magnitude are equally effec
tive in recognizing different textures. Furthermore, a variant of gradient
indexing called local activity spectrum is proposed and shown to have impro
ved performance. Local activity spectrum is employed in an image retrieval
system as the texture statistic. The retrieval system is based on a segment
ation technique employing a distance measure called Sum of Minimum Distance
. This system enables content-based retrieval of database images from templ
ates of arbitrary size. (C) 2000 Academic Press.