Fractal feature and texture analysis

Abstract

AbstractAlthough fractal dimension is popular in computer graphics, it is not yet utilized adequately in image analysis. Thus, this paper presents an example of the application of fractal dimension to image analysis. First, the conventional fractal concept is described and then the possibilities of its application to image processing are discussed. Then a new fractal feature is proposed which can be applied to general images. This is a localized, two‐dimensionalized concept of fractal dimension and is considered to have greater image‐analyzing capability than the conventional simple fractals. Next, these features are measured for texture images and their characters and image‐analyzing capabilities are investigated. The main results are as follows: (1) although the fractal dimension is known as a measure of complexity, its characters differ from the entropy or the variance; (2) the fractal dimension assumes unique value for each texture. Especially, if the newly proposed fractal feature is used, it is possible to recognize texture image to some extent.