Maximum-Likelihood Approach to Image Segmentation by Texture

摘要

This paper addresses the problem of segmenting an image by texture. Specifically, the investigators are concerned with estimating the trajectory of the boundary between the two regions characterized by different two-dimensional autocorrelation functions. An algorithm is developed which is an approximation to a maximum-likelihood estimator based on the assumption of Gaussian random fields with known mean and covariance. Preliminary experimental results are given. (Author)