Texture is a fundamental characteristic in many natural images that, in addition to color, plays an important role in human visual perception and in turn provides information for image understanding and scene interpretation. Multiresolution simultaneous autoregressive models (MSAR) may be viewed as texture features that can be used for image segmentation The MSAR coefficients at different resolution levels are obtained from the respective level of a Gaussian pyramid, and are normalized before clustering them for segmentation. In this paper, we discuss an improved version of MSAR texture segmentation derived by (a) the method used for the construction of the multiresolution image pyramid whose levels are used for calculating the higher level coefficients, and (b) the coefficient normalization process designed to emphasize the information that is most important in the segmentation process.
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
IEEE Western New York Image Processing Workshop 1998
RIT – Main Campus