The analysis of oriented textures through phase portraits

Abstract

A solution for signal-to-symbol transformation in the domain of flowlike or oriented texture is developed. The geometric theory of differential equations is used to derive a symbol set based on the visual appearance of phase portraits. This theory provides the machinery for describing textures both qualitatively and quantitatively. An attractive feature of this symbol set is that it is domain independent and makes no assumptions about the kind of texture that may be present. The computational framework for starting with a given oriented texture and deriving its symbolic representation is provided. It is based on computing the orientation field for the texture and then using a nonlinear least squares technique over successive windows to determine the changing spatial behavior of the texture. The results of applying this technique to several real texture images are presented.