Filtering for texture classification: A comparative study

In this paper, we review most major filtering approaches to texture feature extraction and perform a comparative study. Filtering approaches included are Laws masks, ring/wedge filters, dyadic Gabor filter banks, wavelet tran sforms, wavelet packets and wavelet frames, quadrature mirror filters, disc rete cosine transform, eigenfilters, optimized Gabor filters, linear predic tors, and optimized finite impulse response filters. The features are compu ted as the local energy of the filter responses. The effect of the filterin g is highlighted, keeping the local energy function and the classification algorithm identical for most approaches. For reference, comparisons with tw o classical nonfiltering approaches, cc-occurrence (statistical) and autore gressive (model based) features, are given. We present a ranking of the tes ted approaches based on extensive experiments.