Abstract

Title : Image Compression and Multiscale Approximation
Authors : Albert Cohen and Jacques Froment

This paper deals with image compression problems, by the use of the orthonormal wavelet transform. We calculate the mean square error between an random analog picture and its sampling represented by a partial expansion on an orthonormal basis. We examine the case of several wavelet bases and the Fourier case. The error of compression is also computed in the discrete case using S.Mallat's FWT algorithm (Fast Wavelet Transform). The results of different wavelet bases are compared with Fourier and standard transforms.