Search

In digital signal processing it is often advantageous to analyze a given signal using an adaptive method. The signal is approximated or represented as a superposition of "basic" waveforms chosen from a dictionary of such waveforms so as to best match the signal. The matching pursuit algorithm of Mallat and Zhang is such a method and is discussed in the context of discretized Gabor functions on an interval. We describe two software implementations based on these dictionaries. Both implementations rely on functions defined on an interval to avoid edge effects. One implementation allows for users to have great flexibility in the Gabor dictionary to be used. This is a useful improvement over other implementations, which only allow for a fixed dictionary. The other implementation takes advantage of the FFT algorithm and is faster. These implementations are written in C++, and can be used in practical applications. Categories and Subject Descriptors: G.1.2 [Numerical Analysis]: Approximations; G.4 [Mathematics of Computing]: Mathematical Software; I.4.5 [Image Processing]: Reconstruction General Terms: Algorithms Additional Key Words and Phrases: Adaptive signal processing. Gabor analysis

1. INTRODUCTION The testing and comparative analysis of numerical algorithms is a significant aspect of mathematical software evaluation. Experimental algorithmics deals with the design of methodologies for the consistent evaluation […], A recurring theme in mathematical software evaluation is the generalization of rankings of algorithms on test problems to build knowledge-based recommender systems for algorithm selection. A key issue is to profile algorithms in terms of the qualitative characteristics of benchmark problems. In this methodological note, we adapt a novel all-pairs algorithm for the profiling task; given performance rankings for m algorithms on n problem instances, each described with p features, identify a (minimal) subset of p that is useful for assessing the selective superiority of an algorithm over another, for all pairs of m algorithms. We show how techniques presented in the mathematical software literature are inadequate for such profiling purposes. In conclusion, we also address various statistical issues underlying the effective application of this technique. Categories and Subject Descriptors: G.4 [Mathematics of Computing]: Mathematical Software--Certification and testing General Terms: Algorithms, Performance Additional Key Words and Phrases: Profiling, performance evaluation, benchmark studies, Experimental algorithmics