Mixed noise removal of image with interpolation

Nowadays, natural images in the real world can be affected by more than one type of noise during the process of image acquisition and transmission. Many researchers have been trying to remove the mixed noise because it is a major problem to be considered in image processing. Therefore, they utilized many denoising methods for removal of mixed noise because different noises have different properties. In this paper, we consider the removal of mixed noise composed of Additive White Gaussian Noise (AWGN) and Random-Valued Impulse Noise (RVIN). Although most mixed-noise removal methods can successfully suppress the noise, some image details are lost in edge and texture regions because of the miss-detection of the image details as the impulse noise. Hence, we propose a mixed-noise removal method to preserve the image details. Our method is divided into two steps. The first step is to estimate the denoised image by integrating interpolation, DWM filter, down-sampling and BM3D. The second step is to preserve the image details lost in the first step by calculating the absolute difference between the input noisy image and the pre-estimated image obtained from the first step. The core of this paper is that the input noisy image is initially interpolated by multi-surface fitting for single frame before impulse noise detection of DWM filter in the first step to maintain the image details in the edge and texture regions. Experimental results show that our mixed noise removal method is superior to the state-of-the-art image denoising methods in terms of both quantitative measure and visual perception quality.