A novel image denoising scheme based on fusing multiresolution and spatial filters

The denoising of natural images corrupted by noise is a long established problem in signal or image processing. This paper proposes an effective denoising scheme to remove Gaussian noise by combining spatial filtering and multiresolution techniques. The spatial filter employed here is Joint Bilateral Filter. The Bilateral Filter is a nonlinear filter that does spatial averaging without smoothing edges; it has shown to be an effective image denoising technique. The Joint Bilateral Filter is similar to Bilateral Filter, but it needs a reference image for the parameter estimation. In the proposed scheme, noise-free image is taken as the reference image. The multiresolution techniques applied in this paper are Wavelet Transform, Contourlet Transform and Non-Subsampled Contourlet Transform. In the transformed domain, Bayes thresholding is performed on the detail sub-bands, while Joint Bilateral Filter is applied as the pre-filter and post-filter. The performance is evaluated in terms of Peak Signal to Noise Ratio, Image Quality Index and Edge Keeping Index. The experimental results proved that this algorithm is competitive with other denoising schemes.