Adaptive morphological representation of signals: Polynomial and wavelet methods (Q1361794)

In signal analysis, the concept of filter bank [\textit{P. P. Vaidyanathan}, Multirate systems and filter banks. Prentice Hall, Englewood Cliffs, NJ (1993; Zbl 0784.93096), \textit{M. Vetterli} and \textit{J. Kovačević}, Wavelets and subband coding. Prentice Hall, Englewood Cliffs, NJ (1995), \textit{W. J. Schempp}, Magnetic resonance imaging: Mathematical foundations and applications. Wiley (1998)] is a flexible tool for the decomposition of signals into subbands and their perfect reconstruction procedures. The paper under review proposes a morphological wavelet representation (MWR), analogous to the filter bank technique, that is implemented by an interpolation procedure based on morphological operations [\textit{R. M. Haralick, X. Zhuang, C. Lin} and \textit{J. Lee}, The digital morphological sampling theorem. IEEE Trans. Acoustics Speech Signal Process. 37, 2067-2090 (1989); \textit{I. Pitas} and \textit{A. Venetsanopoulos}, Morphological shape representation. Proc. IEEE Int. Conf. Acoustics Speech Signal Process. 4, 2381-2384 (1991)]. In this way, a pyramide-like structure is derived which allows a signal decomposition into coarse and fine information components, at different scales, similar to the wavelet transformation. Several examples illustrate the application of the MWR to data compression and image processing. A similar hierarchical tree model has been already used in clinical magnetic resonance imaging [\textit{L. P. Panych} and \textit{F. A. Jolesz}, A dynamically adaptive imaging algorithm for wavelet-encoded MRI, Magn. Reason. Med. 32, 738-746 (1994)].