Statistical texture classifcation via histograms of wavelet filtered images

We show how the first order statistic, i.e. the histogram, of the wavelet filtered image is related to the higher order statistic of the original image. We then propose a texture classification method that uses the histogram information of the filtered images as the feature vector. The histogram based features are able to provide richer description of the texture information. Classification experiments show that the proposed method achieves significantly better classification result than the traditional Gabor wavelet filtering method that uses only the mean and standard deviation of the filtered image as feature vectors.