Source Code for Several Fundamental Compressed Sensing (CS) Reconstruction Algorithms

This article presents source code implementations for several fundamental compressed sensing (CS) reconstruction algorithms. The implementations cover MP (Matching Pursuit), OMP (Orthogonal Matching Pursuit), CoSaMP (Compressive Sampling Matching Pursuit), StOMP (Stable Orthogonal Matching Pursuit), and SAMP (Smoothed L0 Orthogonal Matching Pursuit) algorithms. These algorithms are specifically designed to address challenges in signal processing and image reconstruction applications. While these algorithms share common objectives in sparse signal recovery, their implementation approaches differ significantly in terms of atom selection strategies, residual updating mechanisms, and convergence criteria. The MP algorithm employs a simple greedy approach to iteratively select dictionary atoms, while OMP enhances this process through orthogonal projection for improved accuracy. CoSaMP incorporates a pruning mechanism to maintain solution sparsity, and StOMP introduces threshold-based selection for computational efficiency. SAMP implements a smoothed L0-norm approximation with adaptive step size control. Each implementation requires careful analysis of parameter tuning, stopping conditions, and computational complexity to ensure optimal performance in practical applications.