An efficient algorithm for time-domain computation of discrete wavelet transform

Several algorithms are available for efficient computation of Discrete Wavelet Transform (DWT) [1]. Most of them compute sampled coefficients of the continuous wavelet transform using the filter bank structure of the DWT. In this paper we propose a direct method of DWT computation, which is computationally more efficient than the filter bank structure and suitable for parallel processing environment. FFT techniques could be applied here making it still attractive in implementation. The proposed method is found better for smaller input sequences and smaller levels of decomposition. Hence for real time signal processing applications this method could be efficiently employed against the filter bank implementation.