Attitude propagation from digital images

In this paper one introduces an innovative attitude propagation system for remote sensing satellites based on a series of overlapped images taken from a digital camera on the satellite. The system aims to relief the ground control software from the need of identifying landmarks in every payload image and is not based on the expensive and sensitive gyro technology. The attitude propagation procedure starts by fitting a space correlation model to the first image of the series. Then the "kriging" technique is used to predict the next image of the series, as a function of its position and attitude small shifts. The attitude shift is such that the predicted image matches the next taken image. The accuracy of the procedure is evaluated by digital simulation with realistic Earth's surface patterns. Numerical results are presented as a function of the image resolution. They give a preliminary idea about the system feasibility envisaging its experimental implementation in space.