Texture characterisation using a novel optimisation formulation for two-dimensional autoregressive modelling and K-means algorithm

In this paper, texture characterisation is attempted by two-dimensional autoregressive (AR) modelling, which is achieved by using a proposed constrained optimisation technique followed by the application of the k-means algorithm. The constrained optimisation formulation relates both second and third-order statistical parameters of the image with the AR model formulation. The k-means algorithm is applied to sets of AR model coefficients obtained by the proposed method in block-by-block process. A weighting scheme is then used to calculate the final estimated AR model coefficients.