Lognormal random field models and their applications to radar image synthesis

Lognormal random fields with multiplicative spatial interaction are proposed for modeling radar image intensity. A class of two-dimensional (2-D) lognormal random fields, namely the multiplicative Markov random fields (MMRF), is introduced. The MMRF models are formulated as invertible point-transformations of Gaussian Markov random fields (GMRF) and therefore possess many desirable properties. Maximum-likelihood estimates for random field parameters are presented, and techniques for synthesizing 2-D lognormal random fields are discussed. The MMRF models were fit to SEASAT SAR images and then the models were used to generate synthetic images which closely resemble the original SAR images.