摘要:
This paper gives a comprehensive treatment of the PNP (Perspective-N-Point) problem incomputer vision: find the 3-D motion parameters of a viewed object relative to a camera fromthe observed image points of n known feature points. This problem is essentially one of nonlinearoptimization due to the image noise. In this paper, a closed-form solution is developed,and several methods to combat the image noise are presented. The simulation results show thatthe approaches given in this paper hold great promise in application.
展开