A modified approach for noise estimation in optical remotely sensed images with a semivariogram: Principle, simulation, and application

In this paper, a modified approach for noise estimation in optical remotely sensed images is developed under the framework of semivariogram (SV) technique in geostatistics, which is a fundamental and important task for on-ground quantitative application. In comparison with the original method, which involves the extrapolation of modeled SV to the ordinate, in the modified approach, the relationship between two different SVs of true objects at different lags is established. Simulation results show that the latter is more accurate and stable in estimating noise, particularly in the conditions of usual subimage sizes (i.e., 16 x 16 and 32 x 32) as well as in lower noise level. Moreover, the potential negative values in the original method no longer exist in the modified one. Additionally, after the removal of the analog-to-digital conversion noise effects, detection sensitivity evaluations and long-term surveillances for on-orbit optical remotely sensed instrument, for example FY-2 visible infrared spin-scan radiometer, are performed successfully, and the absolute error for noise-equivalent delta temperature estimation is within 0.05 K at 300 K. A common and feasible way for estimating nugget variance of SV in geostatistics is proposed in this paper with the assumption of stationary for both objects and noise process.