A contour matching approach for accurate NOAA-AVHRR image navigation

Abstract

Although different methods for NOAA AVHRR image navigation have already been established, the multitemporal and multi-satellite character of most studies requires automatic and accurate methods for navigation of satellite images. In the proposed method, a simple Kepplerian orbital model for the NOAA satellites is considered as reference model, and mean orbital elements are given as input to the model from ephemeris data. In order to correct the errors caused by these simplifications, errors resulting from inaccuracies in the positioning of the satellite and failures in the satellite internal clock, an automatic global contour matching approach has been adopted. First, the sensed image is preprocessed to obtain a gradient energy map of the reliable areas (sea-land contours) using a cloud detection algorithm and a morphological gradient operator. An initial estimation of the reliable contour positions is automatically obtained. The final positions of the contours are obtained by means of an iterative local minimization procedure that allows a contour to converge on an area of high image energy (edge). Global transformation parameters are estimated based on the initial and final positions of all reliable contour points. Finally, the performance of this approach is assessed using NOAA 14 AVHRR images from different geographic areas.