![Sample our Environment & Agriculture journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5934/TOC-Subject-Sample-2021-Environment-Agriculture.jpg"})
Abstract
Radiometric calibration is the foundation for remote sensors to accurately record the reflected energy from targets and to also effectively display the reflectance diversity among them. As one of the calibration methods, pre-launch laboratory relative calibration is essentially a normalizing process for each detector of a sensor at different intensity levels of various radiation sources. However, interferences such as stray light, dark current, and stochastic noise will cause some deviation of the normalizing correction factor. In this article, we propose an integral noise (a combination of the aforementioned three noises) estimation method based on the correlation between the elements of the calibration data itself. Abbreviated as IMRAN (Iterative Maximal Residual As Noise), this method is an iteration procedure using least square fitting to calculate the maximum residual of the sensor pixel in question against the rest sensor pixels and to consider this value as the estimated noise. The iteration is continued after subtracting the noise from the raw data of the sensor pixel until the noise estimation gets converged and then the accumulation of the results from each round is the final estimated noise. And this procedure is applied to every sensor pixel. The verification results demonstrated the IMRAN method can effectively estimate the integral noise of pre-launch radiometric calibration data and substantially improve its precision. When the number of radiation level increases, the precision of the estimated noise will be rapidly increased, whereas the number of sensor pixels has no obvious effect. Because this IMRAN method uses the data of every sensor pixel, it is sensitive to the outlier, which can be eliminated by variance detection as part of the IMRAN method.