Real-time wavelet transform with overlap of signal segments

The processing of various types of signal is often required to be in ldquoreal timerdquo. The recently developed method of segmented wavelet transform (SegWT) makes it possible to compute the discrete-time wavelet transform of a signal segment-by-segment, not using the traditional ldquowindowingrdquo. The method can be utilized for wavelet-type processing of a signal in real time or in case we want to process the signal in parallel or in case we need to process a long signal, but available memory capacity is insufficient (e.g. in the DSPs). In the paper, the principle of both the forward and the inverse SegWT is explained and the algorithm integrating them in the real-time case is given.