Additive Noise Level Estimation Based on Singular Value Decomposition (SVD) in Natural Digital Images

• Published in: 2019 IEEE International Conference on Signal and Image Processing Applications (ICSIPA) pp. 225 - 230
• Main Authors: Khmag, Asem, Malallah, Fahad Layth, Sharef, Baraa T.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2019
• Subjects: additive white Gaussian noise; image denoising; Noise estimation; singular value decomposition; visual quality
• ISSN: 2642-6471
• DOI: 10.1109/ICSIPA45851.2019.8977764

True Noise level estimation is a seminal research of interest in the topic of digital image processing especially in blind noise removal methods. In this study, an estimation of additive white Gaussian noise (AWGN) in digital natural images is introduced. The adaptive noise level estimation is designed mainly based on singular value decomposition (SVD) of the natural images. The proposed technique contains two pivotal stages. Firstly, typical noise level estimate is utilized in order to manipulate the algorithm factors to be used in the second stage of the proposed technique. Secondly, the adjusted parameters are used in SVD in order to speed up the estimation processes and increase the accuracy rate of the noise level estimation. The experimental results depict that the proposed algorithm performs professionally over a several ranges of visual details which is presented in PSNR and MSE in AWGN removal methods. In addition, in terms of time complexity, the proposed algorithm in second stage shows significant performance in terms of computational load and achieves high running speed.