Fast Noise Level Estimation using a Convergent Multiframe Approach
Angelo Bosco, Arcangelo Ranieri Bruna, Stewart Smith, Valeria Tomaselli · 2006
Image denoising is one of the most recurrent problems to face in the design of image generation pipelines. This paper proposes a novel method for the estimation of the noise level in images contaminated by additive white Gaussian noise (AWGN). The estimated noise level is updated on a frame basis allowing convergence to the optimal value. At the same time, dynamic adjustment to changing noise levels is permitted. The proposed technique has been successfully implemented in a real system, demonstrating the validity of the proposed solution.