Contourlet domain image denoising based on the Bessel k-form distribution

Statistical image modeling has attracted great attention in the field of image denoising. In this work, a new image denoising method in the contourlet domain is introduced in which the contourlet coefficients of images are modeled by using the Bessel k-form prior. A noisy image is decomposed into a low frequency approximation sub-image and a series of high frequency detail sub-images at different scales and directions via the contourlet transform. To estimate the noise-free coefficients in detail subbands, a Bayesian estimator is developed utilizing the Bessel k-form distribution. In order to estimate the parameters of the distribution, a characteristic function-based technique is used. Simulation results on standard test images show improved performance both in visual quality and in terms of the peak signal-to-noise ratio and structural similarity index as compared to some of the existing denoising methods. The proposed method also achieves an excellent balance between noise suppression and details preservation.