I wanted to classify a satellite image by spectral discrimination, but it became obvious that spectral analysis is not only sufisante. Discrimination by texture analysis must complete the protocol (as shown in the development of expensive software that offer this approach as eCognition software). I found Haralick texture index seems to be the most used. I also found the literature on texture index calculated by wavelets. What is the added value of a method over the other? (Haralick / wavelets)