Scale limits in estimating fractal dimensions of image surfaces

Fractal dimension is a useful parameter for characterizing textured images. However,different methods and even the same method with different scale ranges of computation give different values due to the finite sizes and gray level numbers of digitized images. This paper presents an intuitive, simple, model-free algorithm for estimating the fractal dimensions of textured images. The emphasis is on how scale range affects the value of the measured fractal dimension. The results of experiments using computer-generated image surfaces demonstrate the validity of the scale limit criterion.