Proc. SPIE. 8768, International Conference on Graphic and Image Processing (ICGIP 2012)
KEYWORDS: Signal to noise ratio, Digital signal processing, Chemical species, Interference (communication), Signal processing, Reconstruction algorithms, Electronic filtering, Detection theory, Filtering (signal processing), Algorithms
Compressed Sensing (CS) that can effectively extract the information contained in the signal is a new sampling theory based on signal sparseness. This paper applies CS theory in digital speech signal enhancement processing, proposes an adaptive quantile method for the noise power estimation and combines the improved double-threshold orthogonal matching pursuit algorithm for speech reconstruction, then achieves speech enhancement processing. Compared with the simulation results of the spectral subtraction and the subspace algorithm, the experiment results verify the feasibility and effectiveness of the algorithm proposed in this paper applied to speech enhancement processing.