De-noising Raman Spectra using a Non-convex Regularizer in Total Variation

Objectives: This article deals with a novel de-noising method for Raman Spectra based on total variation de-noising in which a non-convex regular is used. Methods/Statistical analysis: Total variation de-noising is expressed as an optimization problem with a quadratic data fidelity term and a non-convex regularizer maintaining the convexity of the problem. Problem is solved using two different non-convex regularizers: logarithmic and arctangent. Performance of proposed methodology is evaluated by finding the Signal to Noise Ratio (SNR) and Root Mean Square Error (RMSE). Also, result is compared with de-noising using convex regularizer. Findings: From the calculated SNR and RMSE, it is observed that the proposed method works well and produces better results than the result obtained by the method using the convex regularizer in total variation de-noising. Application/Improvements: The proposed de-noising method can be used for other signal processing applications.