Unsupervised texture segmentation using discrete wavelet frames

Image segmentation could be based on texture features. In this work, an unsupervised algorithm for texture segmentation is presented. Texture analysis and characterization are obtained by appropriate frequency decomposition based on the Discrete Wavelet Frames (DWF) analysis. Texture is then characterized by the variance of the wavelet coefficients. The unsupervised algorithm determines the regions to characterize each different texture content in the image. For applying the algorithm, it is necessary to know only the number of the different texture contents of the image. Then, based on a distance measure, each point of the image is classified to one of the different contents.