Mixed Gaussian-impulse noise reduction from images using convolutional neural network

The removal of mixed-noise is an ill-posed problem due to high level of non-linearity in the distribution of noise. Most commonly encountered mixed-noise is the combination of additive white Gaussian noise (AWGN) and impulse noise (IN) that have contrasting characteristics. A number of methods from the cascade of IN and AWGN reduction to the state-of-the-art sparse representation have been reported to reduce this common form of mixed-noise. In this paper, a new learning-based algorithm using the convolutional neural network (CNN) model is proposed to reduce the mixed Gaussian-impulse noise from images. The proposed CNN model adopts computationally efficient transfer learning approach to obtain an end-to-end map from noisy image to noise-free image. The model has a small structure yet it is capable of providing performance superior to that of the well established methods. Experimental results on different settings of mixed-noise show that the proposed CNN-based denoising method performs significantly better than the sparse representation and patch-based methods do both in terms of accuracy and robustness. Moreover, due to the lightweight structure, the denoising operation of the proposed CNN-based method is computationally faster than that of the previously reported methods. •A novel CNN-based method for reducing mixed Gaussian-impulse noise from images.•New mapping from noisy to noise-free images using a 4-stage CNN architecture.•Adoption of transfer learning for faster training of proposed CNN model.•Experiments on challenging datasets with diverse settings of noise parameters.•Results show that proposed method is better than existing or similar methods.