Greedy algorithms and \(M\)-term approximation with regard to redundant dictionaries (Q1288881)

The author considers a particular method of nonlinear approximation, namely \(M\)-term approximation; it is used in image and signal processing. He studies the efficiency of greedy type algorithms with regard to redundant dictionaries in Hilbert space and proves a general result which gives a sufficient condition on a dictionary to guarantee that the pure greedy algorithm is nearly best in the sense of power decay of error of approximation. Some examples are given.