A novel dual-threshold denoising method by the wavelet transform

Currently image denoising uses various threshold methods based on wavelet transform. In this paper, we propose an effective and novel dual-threshold denoising method that uses a piecewise and nonlinear function and selects two thresholds for every image subband, noise threshold T_n and signal threshold T_s. Then we further study the subband coefficients: those which are lower than T_n are noises and will be removed, those which are higher than T_s are signals and will be retained and those which are between T_n and T_s will be handled by their “intensity factors”. The method powerfully distinguishes the noise and the signal, overcomes the limitations of the traditional methods of the wavelet denoising using single threshold to deal with the coefficients. Experiments show that the method is better than other traditional ones in terms of SNR and visual effects.