Image segmentation and contour detection using fractal coding

Fractal coding was applied to image segmentation and contour detection. The encoding method was the same as in conventional fractal coding, and the compressed code, which we call the fractal code, was used for image segmentation and contour detection instead of image reconstruction. An image can be segmented by calculating the basin of attraction on a mapping that is a set of local maps from the domain block to the range block. The local maps are parameterized using the fractal code, and contours of the objects in the image are detected by the inverse mapping from the range block to the domain block. Some objects in the test image Lena were segmented, and the contours were detected well. The proposed methods are expected to enable compressed codes to be used directly for image processing.