Image noise variance estimation using the mixed Lagrange time-delay autoregressive model |
| |
Authors: | Sim K-S Tso C-P Law K-K |
| |
Affiliation: | Faculty of Engineering and Technology, Multimedia University, Melaka, Malaysia. sksbg2003@yahoo.com |
| |
Abstract: | The mixed Lagrange time-delay estimation autoregressive (MLTDEAR) model is proposed as a solution to estimate image noise variance. The only information available to the proposed estimator is a corrupted image and the nature of additive white noise. The image autocorrelation function is calculated and used to obtain the MLTDEAR model coefficients; the relationship between the MLTDEAR and linear prediction models is utilized to estimate the model coefficients. The forward-backward prediction is then used to obtain the predictor coefficients; the MLTDEAR model coefficients and prior samples of zero-offset autocorrelation values are next used to predict the power of the noise-free image. Furthermore, the fundamental performance limit of the signal and noise estimation, as derived from the Cramer-Rao inequality, is presented. |
| |
Keywords: | Cramer–Rao lower bound noise variance noise signal |
本文献已被 PubMed 等数据库收录! |
|