Metric planar rectification from perspective view via circles

Abstract

In this paper, we show that the images of the circular points can be identified by solving the intersection of two imaged coplanar circles under projective transformation and thus metric planar rectification can be achieved. The advantage of this approach is that it eliminates the troublesome camera calibration or vanishing line identification step that underlies many previous approaches and makes the computation more direct and efficient. Vanishing line identification becomes a by-product of our method. Different root configurations are inspected to estimate the image of the circular points reliably so that 2D Euclidean measurement can be directly made in the perspective view. The experimental results validate the effectiveness and accuracy of the method.