Projective invariant based crater matching for visual navigation in planetary landing

Abstract

This paper presents a new crater matching algorithm for visual navigation in planetary landing. For pin-point landing mission, IMU could not satisfy its goal because of cumulated error. Therefore crater detection and matching algorithm(CDMA) have been developed on that account. Since crater is a landmark on planetary surface, information about its absolute position can be obtained in advance. If a crater successfully detected and matched from the camera image, its absolute position helps to determine the landers position that can be used in navigation for pin-point landing mission. This paper concentrates only crater matching not detection. Projective invariant feature with five coplanar points are used to match crater. On crater image 4 sets of five coplanar points are obtained from intersection and cross-intersection between two ellipses and 2 invariants can be calculated according to each set of point respectively. To match image crater with database crater simple voting algorithm is used. LU78287GT, Lunar impact crater database, is used to simulate an image. Finally, performance of proposed crater matching algorithm will be compared with varying noise condition.