Study on Total Variational Denoising Algorithm Based on Penalty Term of Exponential Function

To solve the problem of signals denoising with continuous and large jump of peak values, a new penalty term based on total variation denoising algorithm is proposed in this paper. The exponential function is used to exert the penalty property for better restrain the noise with large amplitude. The effect of denoising is improved by introducing the exponential function into the cost function. The new penalty term, which could make the denoising signal smoother and the disturbance fluctuation smaller, can effectively reduce the noise of the interference signal with discontinuous point. The experimental results show that, compared with the optimal minimum algorithm, the proposed method improves the denoising accuracy of the signal with fluctuation and peak jump in a large range.