Sensor Noise Assessment Approach in a Patterned Region

Abstract

In this paper, we propose a simple noise assessment approach for sensors. We assume the input signal is contaminated by white additive and zero mean Gaussian noise which. To measure the noise level, we the intensity-homogeneous blocks are found first, and then we evaluate the noise variance in the blocks. To determined intensity-homogeneous blocks, we use high pass filter. Experimental results show that the presented method provides good performance for natural and artificial images over a large range of noise levels.