Signal enhancement and the method of successive projections (Q1177139)

The signal enhancement algorithm developed by \textit{J. A. Cadzow} in a series of papers [e.g. IEEE Trans. Acoust. Speech Signal Process. ASSP-- 36, No. 1, 49-62 (1988; Zbl 0649.93059)] treats the problem of finding a signal which is known to process certain a-priori properties and that lies closest to a measured signal. The measured signal itself often fails to have one or more of these properties, e.g. due to measurement noise. In the present paper this signal enhancement algorithm is introduced on the basis of the method of successive projections (MOSP) which was investigated by \textit{P. L. Combettes} in his PhD thesis (1989). This method is a generalization of the method of projections on convex sets (POCS). The necessary definitions, theorems, and corollaries are summarized. A couple of examples for the application of the algorithm is presented, including: MUSIC array processing, direction-of-arrival for multiple broadband incoherent sources, data interpolation of exponential signals, detection of sinusoids in white noise, and \(x\)-ray flourescence spectra. For further reading this review article is supplied with about 60 references.