Entropy measures for texture analysis based on Hadamard transform

The implementation of a software texture analysis technique for images based on the entropy of transform components is described. Different images were compared by computing their entropies, resulting in a relative texture measure. Since entropy can be based on the statistics of both gray-level components and frequency components, four entropy computation routines were implemented. The software calculates the entropy of the gray-levels histogram, the frequency components using the fast-Fourier transform (FFT), transform components using the discrete-cosine transform (DCT), and the transform coefficients of the fast-Hadamard transform (FHT). Because of its decrease in computation time and storage requirements, fast Hadamard was chosen as a better transformation. Entropy calculations were tested on different regional sizes in a time-varying imaging system in order to study regional variations of texture