Multifractal wavelet compression of fingerprints

This paper presents compression of grey-scale fingerprint images, using a wavelet transform guided by a multifractal measure to obtain the best reconstructed image in terms of a higher peak signal to noise ratio, PSNR, at the lowest bit rate. The fingerprint images and the corresponding wavelet coefficients are considered to be approximation of strange attractors and can be analyzed by their multifractality. The wavelet can provide not only the grouping of subbands information and the highest compression for optimum bit allocation (quantization), but also an optimum synthesis (combination of subbands) by the inverse wavelet transform to achieve the highest image quality. The motivation for this paper is to find the best combination of the subbands for both the quantization and image quality by applying the Mandelbrot (1983) singularity measure to the coefficients in various subbands