Satellite sensor modeling and 3D geo-positioning using empirical models

The need for the use of general empirical mathematical models for satellite sensor modeling and 3D geo-positioning has increased recently, mainly because of the absence of the satellite sensor information of some of the high-resolution satellites. In addition, empirical mathematical models can be applied to different satellite sensors since they are time independent mathematical models and do not require specialized commercial software packages. This paper discusses the applicability of the empirical mathematical models presented by the 3D affine model and the 3D polynomial models for satellite sensor modeling and 3D geo-positioning. The objectives of the paper are to demonstrate that (a) the 3D affine model and its modifications of the 3D polynomial models are applicable to different satellite sensors and different types of terrain, and (b) under some conditions, the empirical models can produce accuracies close to those from rigorous mathematical models. data sets are used in this work from three different satellite sensors (SPOT 4, IRS/1D, and Ikonos) covering different types of terrain (flat, hilly, and mountainous terrain). Conditions and limitations of the models and the data sets used are discussed in the experimental work. The results reveal that the empirical mathematical models can be used successfully with different types of satellite sensors and for different types of terrain. Generally, one-to-three pixels accuracy is achievable when a modest number of accurate Ground Control Points is used.