Nonparametric multichannel texture description with simple spatial operators

A multichannel approach to texture description is proposed by approximating joint occurrences of multiple features with marginal distributions, as 1D histograms, and combining similarity scores for 1D histograms into an aggregate similarity score. A stepwise feature selection algorithm is used to choose the best feature combination in a particular dimension. In classification experiments with Brodatz textures and MeasTex test suites the proposed method performs favorably compared to GLCM, Gabor and GMRF features.