RANSAC-Based Signal Denoising Using Compressive Sensing

In this paper, we present an approach to the reconstruction of signals exhibiting sparsity in a transformation domain, having some heavily disturbed samples. This sparsity-driven signal recovery exploits carefully suited random sampling consensus (RANSAC) methodology for selection subset inlier To aim, two fundamental properties are used: A sample represents linear combination sparse coefficients, whereas disturbance degrades original sparsity. The properly selected samples further used as measurements reconstruction, performed using algorithms from compressive sensing framework. Besides fact that no other disturbance-related assumptions made—there special requirements regarding its statistical behavior or range values. As case study, discrete Fourier transform is considered domain sparsity, owing significance processing theory and applications. Numerical results strongly support presented theory. addition, exact relation signal-to-noise ratio reconstructed also presented. simple result, which conveniently characterizes RANSAC-based performance, numerically confirmed by set examples.