Knowledge-enhanced Matching Pursuit

Compressive Sensing is possible when the sensing matrix acts as a near isometry on signals of interest that can be sparsely or compressively represented. The attraction of greedy algorithms such as Orthogonal Matching Pursuit is their simplicity. However they fail to take advantage of both the structure of the sensing matrix and any prior information about the sparse signal. This paper introduces an oblique projector to matching pursuit algorithms to enhance detection of a component that is present in the signal by reducing interference from other candidate components based on prior information about the signal as well as the structure of the sensing matrix. Numerical examples demonstrate that performance as a function of SNR is superior to conventional matching pursuit.