Adaptive Thresholding Denoising Algorithm Based on Cross-Validation

In this paper, a novel wavelet-based adaptive thresholding de-noising algorithm is proposed. By using a modified twofold cross-validation, a noise-corrupted signal is divided into two parts: one for estimating, the other one acts as a reference signal, and they make it possible to search for the optimal threshold using steepest gradient method. The numerical results indicate that the proposed optimal-threshold-based denoising algorithm outperforms the standard wavelet shrinkage methods, like Donoho´s VisuShrink and SureShrink, in MSE sense. The proposed algorithm does not need any a priori information of the noise-distorted signal, and its convergence speed is high. It fits to real-time signal processing