On the construction of joint probabilistic image models

A method is proposed for constructing the joint probability model from a specified first-order distribution and correlation structure. The approach is based on an invertible nonlinear transformation. The model's information-theoretic properties (entropy, rate-distortion bound) are examined. The results can be used to analyze and evaluate the performance of an image data compression system. It is believed that this class of model and its information-theoretic properties may make for more realistic modeling of medical images and hence their more efficient processing.<>