Noise bias compensation based on Bayesian inference for tone mapped noisy image

Abstract

This paper introduces a noise bias compensation to a tone mapped noisy image so that the variance of the noise is reduced. Although the noise bias is assumed to be zero before tone mapping (TM), it becomes non-zero value after TM. The reason includes some factors such as the non-linearity of TM and the asymmetry of the probability density function of the noise. In this paper, pixels in the noisy image are classified into several subsets according to the observed pixel value, and compensates the pixel value in each subset with a preliminary determined compensation value (CV). In this paper, CV is determined from the histogram of pixel values in the image and that of the noise before TM or their modeled versions based on the Bayesian inference deterministically. As a result of experiments, it is observed that the peak-signal to noise ratio is improved by the proposed method.