Corner Feature Extraction and Matching Method Constrained by Edge Region for Lunar-based Equipment Optical Navigation

Aiming at the unstructured environment of the lunar surface, and the problems such as gray scale changes and texture features are not obvious, a method for extracting and matching optical navigation features of lunar-based equipment under edge region constraints is proposed. By setting double thresholds, weak edge pixels are suppressed to screen out edge information, an edge information domain is constructed, and accelerated robust feature descriptors of edge images are extracted and matched. The search range of corner detection is reduced effectively and the concentration of feature detection is improved. Then we use the estimator sample consistency algorithm to eliminate the wrong matching, and finally achieve accurate matching while preserving the feature points of the lunar surface image.