Randomized continuous frames in time-frequency analysis (Q2135622)

The authors present and analyze applications of a Monte Carlo method for highly redundant continuous frames. The computational complexity of the method depends on the linear volume discretize property (LVD), which means that the number of samples in the coefficient space required by the Monte Carlo method is proportional to the resolution of the discrete signal. The authors show that the continuous wavelet transform (CWT) and the localizing time-frequency transform (LTFT) have the LVD property. The LTFT is a time-frequency representation with a wider class of time-frequency atoms than the classical time-frequency transform. The complexity of the LTFT Monte Carlo method is proved to be asymptotically equivalent to the complexity of the CWT and the short-time Fourier transform methods.