This study focuses on an application of the Orthogonal Matching Pursuit (OMP) technique for interpolation and denoising of seismic signals.OMP is an algorithm for sparse coding based on orthogonal projections of the signal over an overcomplete dictionary. This overcomplete dictionary was designed with K-times Singular Values Decomposition (K-SVD). Developed algorithms were applied to VSP seismic data and results achieved, in restored lost traces and denoised signals, are presented.