Texture analysis based on a family of stochastic models

A statistical approach, based on a family of Full Range Autoregressive models, is proposed for texture analysis. Bayesian approach is adopted to estimate the model parameters. Using the parameters, autocorrelation (AC) coefficient is computed. Based on the AC, two texture descriptors are proposed: (i) texnum, the local descriptor and (ii) texspectrum, the global descriptor. Decimal numbers are computed and that are proposed to represent textures present in a small image region. These numbers uniquely represent the texture primitives. The textured image under analysis is represented globally by observing the frequency of occurrences of the texnums called texspectrum. The textures are identified and are distinguished from the untextured regions with edges.