BM3D Image Denoising using Learning-based Adaptive Hard Thresholding

Image denoising is an important pre-processing step in most imaging applications. Block Matching and 3D Filtering (BM3D) is considered to be the current state-of-art algorithm for additive image denoising. But this algorithm uses a fixed hard thresholding scheme to attenuate noise from a 3D block. Experiments show that this fixed hard thresholding deteriorates the performance of BM3D because it does not consider the context of corresponding blocks. In this thesis, we propose a learning based adaptive hard thresholding method to solve this issue. Also, BM3D algorithm requires as an input the value of the noise level in the input image. But in real life it is not practical to pass as an input such noise level. In this thesis, we also attempt to automatically estimate the level of the noise in the input image. Experimental results demonstrate that our proposed algorithm outperforms BM3D in both objective and subjective fidelity criteria.