Fractal Characterization of Irregularly Shaped Images by Frequency Analysis

An iterative spectral method is presented for fractal characterization of irregularly shaped (i.e., nonrectangular) images such as microscopic images of cells. The algorithm extends the application of fractal characterization by frequency analysis to images with irregular supports by utilizing information in both the spatial and frequency domains. Experiments with synthetic fractal images demonstrate the efficacy of the proposed method.