Comparison of two fast algorithms for CWT computations

Continuous wavelet transform (CWT) is a useful technique to analyse time-varying signals. Direct computation of CWT via FFT requires O(Nlog/sub 2/N) operations per scale, where N is the data length. This paper compares two fast algorithms that compute CWT at a cost of O(N) per scale. One is a trous algorithm and the other is Shensa algorithm. Although both are based on the multiresolution analysis structure, their accuracy in computing CWT is quite different. Theoretical error expression is derived and simulation results are presented for comparison.