Image Denoising Method with Adaptive Bayes Threshold in Nonsubsampled Contourlet Domain

In this paper, an improved image denoising method based on a nonlinear thresholding function with adaptive bayes threshold in Nonsubsampled Contourlet Transform (NSCT) domain. In overcoming the shortcomings of the same threshold, the noise deviation of the different sub-band are estimated based on the coefficients of different directional and level sub-bands in NSCT domain, and the thresholds of every sub-band is estimated by Bayesian threshold estimation method. After choosing the thresholds, a nonlinear thresholding function was chosen to overcome the shortcomings of the soft and the hard thresholding function. The simulation results show that the proposed method in this paper can remove Gaussian white noise more effectively, and get a higher PSNR value and keep image texture and detail information more clearly, which also has a better visual effect.