Abstract
Texture as an image contents is fundamental descriptor to represent an image. In this paper, we propose a texture featuring technique and a texture descriptor based on human visual system(HVS). Psychophysical experiments have shown that the response of visual cortex is turned to a band-limited portion of the frequency domain. It gives evidence that the brain decomposes the spectra into perceptual channels that are bands in spatial frequency. As far as spatial vision is concerned, best-adopted sub-band representation of HVS is a division of spatial frequency domain with octave-band division in radial axis and equal-width division in orientation axis(or angular axis). In this paper, we describe the texture in Radon space that is believed to fit HVS.
