Joint sparse representation and denoising method for Raman spectrum

A new method based on joint sparse representation is developed to recover the peak information from high-noise Raman signal. This method used the sparsity of Raman spectrum to recover the signals and preserve its useful peak information. The peak information is then reconstructed by using an orthogonal matching pursuit algorithm. The joint sparse representation method is found to be an effective approach to analyze the Raman spectrum, especially Raman spectrum that have high noise, thus improving the detection limit of Raman spectroscopy. Experimental results demonstrate that this approach is better than other approaches in case of low signal-to-noise ratios.