Study on Vision-Based Navigation-Integration of Coplanarity and Collinearity Condition for Ego-Motion Estamation

摘要

Photogrammetry research generally uses the collinearity condition to establish relations between object and image points and calculate absolute orientation parameters Given that the collinearity condition is a nonlinear system appropriate initial values of unknown absolute orientation parameters must be set up first in the iterative least-squares solution This research proposes using the coplanarity condition to solve relative orientation parameters as well as using these parameters as initial values to solve absolute orientation parameters based on the collinearity condition The proposed method can provide a strategy for solving the absolute orientation parameters of a camera This research focuses on the motion estimation of a single camera First the algorithm of feature point detection is matched to automatically acquire conjugate image points between sequential images Information on conjugate image points can provide observations to solve relative orientation parameters given the coplanarity condition Second results from the previous step can provide initial values and observations The absolute orientation parameters of a camera can be solved through the iterative least-squares method with the aid of ground control points Aside from the absolute orientation parameters of a camera the object-space coordinates of conjugate image points can also be acquired These conjugate image points can be used as new ground control points for subsequent image pairs Finally camera trajectory can be obtained by repeating the procedure This research conducted experiments in indoor and outdoor environments and the results proved that the proposed procedure is effective and feasible

獎項日期	2014 9月 2
原文	English
監督員	Hone-Jay Chu (Supervisor)