Kriging-based technique for remote sensing image restoration

The objective of this paper is to examine the effect of restoration of remotely sensed images by three geostatistical approaches, Ordinary Kriging (OK), Universal Kriging (UT), and Indicator Kriging (IK).Using the undersampled data of NDVI from NOAA/AVHRR image, we obtained OK and UK estimates and E-type estimates of IK, OK and UK variances and conditional variance of IK. After comparison, we found that the images of OK and UK estimates can both successfully restore the overall trend of the original image, but the image of E-type estimates of IK is not good enough. We also found that the images of OK and UK variances only reflect the sampling configuration because OK and UK variances are independent of the data values locally, however, owing to the fact that conditional variances of IK are conditional on the data values, they show the errors of estimates perfectly, and the magnitudes of conditional variances are consistent with the uncertainty of remotely sensed data.