Abstract
Fractal image coding is a block-based scheme that exploits the self-similarity hiding with an image. Fractal codes are quantitative measurements of the self-similarity of the image, and collage error distribution of block characterizes the degree of self-similarity in it. Furthermore, fractal codes can be used to obtain a practical image indexing system because of its compactness and stability. The most important reason using fractal codes is able to deal with the images in compressed form. Thus fractal indexing is suitable for use with large database. In this study, we propose a new image retrieval system and object detection method based on fractal coding features that are collage error distribution and block partition structure in fractal codes. Experimental results show that the proposed method achieves a high precision tracking which is faster than MPEG method.
