Face recognition based on fractal dimension

An approach for face recognition was proposed based on features called fractal dimension in this paper; it is an efficient representation of natural texture. A grayscale image can be divided into many non-overlapping blocks and the fractal dimension can be computed in every single block to keep better performance of representing subtle texture with fractal dimension when it was applied to present facial image. Experiments on ORL face database displayed promising results and they demonstrated that the size of block affect recognition rate.