ONLINE COMPUTATION OF EXTERIOR ORIENTATION WITH APPLICATION TO HAND-EYE CALIBRATION

Computation of the relative position and orientation between a camera and an observed object from a single image is a central problem in com puter vision. Although many solution methods have been proposed, sever al problems remain. Analytical methods do not take into account the is sue of noise. Nonlinear least-squares methods depend critically on goo d initialization. Linear least-squares methods tend to be very sensiti ve to noise and outliers. These shortcomings limit their use in modern computer vision applications. In this article, we formulate a new lea st squares objective function that leads to a good initialization sche me based on weak-perspective projection, as well as a robust and effic ient descent algorithm using absolute orientation. The new method comb ines model-based parameter search and data-driven backprojection which , unlike most existing methods, minimizes 3-D object space error rathe r than 2-D image error. Extensive experiments on simulated data indica te that the new method outperforms commonly used least squares methods under most conditions, Its performance as a kernel in the inner loop of a robust M-estimate algorithm for outlier rejections is also studie d. We demonstrate the use of this method in the context of hand-eye ca libration.