Automatic Recognition and Detection of Lunar Concave Obstacles Based on Shadow Feature

Abstract

The widespread impact craters and other concave obstacles on the lunar surface are the key factors threatening the safe landing and roving of the lunar rover. Once trapped, it will bring risks of tilt, landslide, and even rollover to the lunar rover. Therefore, the effective recognition and detection of lunar concave obstacles are conductive to obstacle avoidance, and provide necessary information reference for the safe landing and roving of the lunar rover. Based on the concave obstacles’ feature that there is a one-to-one matching between the shadows and the highlights in the sun, an automatic recognition and detection method for the lunar concave obstacles is proposed. The adaptive dual threshold method is used to automatically separate the shadows and the highlights of the concave obstacles from the background. Each shadow and highlight are clustered the specific position and one-to-one matched using the sunlight direction with the prior forecast information involved. Then the rough extraction of every single concave obstacle are obtained. Finally the original sub-images sequence containing every single concave obstacle is traversed for edge detection and ellipse fitting, which can avoid mutual interference of multiple obstacles and effectively detect the locations and ranges of all concave obstacles.

Keywords

concave obstacles / threshold segmentation / cluster analysis / shadow matching / edge detection