Fractal segmentation using multiresolution B-spline wavelets

This paper presents a new method for obtaining high-resolution edges of a natural image. The algorithm is based on the estimation of the local fractal dimension (FD) of the image using its local spectral density computed from the Fourier-wavelet transform. Using our high-resolution FD map, the FD and edge information is integrated into a region growing algorithm for segmentation of natural and real images. The efficiency of the method is illustrated on five real images.