Automatic Feature Extraction of Pose-measuring System Based on Geometric Invariants

摘要

This paper proposes a method for extracting features of target object automatically in a pose-measuring system. A tag stuck to the target object is designed to make the system more robust and precise. First, detect all the corners of an image. Then, geometric invariants, which are formed by several invariants of four collinear points,are used to distinguished corners of the tag from all the corners. To alleviate the computational complexity,polar coordinate system is introduced for searching collinear points. Experimental results show that this method can extract features of the target automatically,rapidly and exactly under different viewpoints.