Design and Development of Technique and Software for Vision Based Landmark Detection and Matching

Abstract

Real-time position estimation for a descent lander is a critical technological need for many planned missions. In particular, it enables the ability to land precisely and safely in a scienti cally promising but hazardous site. The primary question of pinpoint landing is how to localize the lander by recognizing landmarks (craters) in the landing area and match them positively to a preexisting landmark database while the spacecraft is descending. Because the crater appearances viewed from orbit and lander are di erent since the imaging conditions are di erent. In that case crater matching becomes di cult. The main objective of this research work is to nd the position of spacecraft us- ing landmarks. Here landmark detection algorithm uses canny edge detection algorithm having adaptive thresholds. So that it can withstand di erent imaging condition. Af- ter detecting landmarks, parameters of landmarks are given as an input to landmark matching algorithm. Landmarks are matched using geometric based method. It requires only relative landmark locations. As it nds unique geometry, it can operate under wide range of crater shadowing, scale uncertainty. Once landmarks are matched with database, matched landmark converted from camera frame to reference frame and position of space- craft referenced from safe landing site is calculated.