Wavelet filter design for image compression

A new approach to wavelet filters optimized for image compression is proposed. Since the dyadic wavelet transform corresponds to convolutions of a signal with a high-pass and a low-pass filter in the time domain, it is useful to design the filters in the time domain aiming at a high energy compaction. The design technique introduced is based on a signal decomposition into suitable base functions. It is shown that this approach can lead to well-known wavelets, and furthermore the construction of new wavelet functions is possible.