Joint analysis of visible/near-infrared and thermal infrared images of ground terrain and the influence of scene topography on the observed joint radiant statistics

Abstract

A physical model for the prediction of the radiant statistics over thermal images of ground desert terrain landscapes and their temporal behavior had been fully established by Ben Yosef et al. This model can be further developed in order to formulate the joint radiant statistics of reflective and thermal infrared images over the same type of landscapes. However, it fails to predict the actual measured correlation between the images in the two bands, and hence, a modification of the joint radiant density function in order to consider the influence of the local ground topography over the scene is introduced. The prediction of the modified joint density function and correlation coefficient is consistent with the experimental data acquired over a rough desert landscape. The effect of local scene topography on thermal image properties is not negligible, especially when sun's elevation angle is low. In such cases, shaded areas are generated in the scene, occupying a substantial portion of it. Analysis of the temporal dynamics of the correlation coefficient between thermal and reflective images can infer about the relative importance of the topography contributed variance over the thermal image and as well as the clutter characteristics of the thermal image. Scene topography also introduces errors in the production of thermal inertia maps for remote sensing applications and limits its utility.