| Hi-index | 0.00 |
In this paper, we propose an orthogonal polynomialsbased color texture model for color texture segmentation.The proposed model is based on a completeset of difference operators which are easily configurablefrom a set of orthogonal polynomials.The operatorsare employed to represent a color image regionas a linear combination of the operator's responses towardstexture and noise.A simple statistical designof experiments paradigm is used for separating outthe responses towards color texture from the responsestowards noise.A local descriptor for color texturecalled cpronum is computed by encoding only the significantresponses towards texture.Finally a globaldescriptor for color texture called cprospectrum is obtainedby measuring the frequency of occurrences ofthe cpronums in the color image.The usage of theproposed color texture representation scheme is highlightedin unsupervised color texture segmentation.