A Multi-frame Structure-from-Motion Algorithm under Perspective Projection

We present a fast, robust algorithm for multi--frame structure from motion from point features which works for general motion and large perspective effects. Experimental results on synthetic and real sequences show that the algorithm gives results essentially as accurate as the maximum likelihood estimate (MLE) in a couple of seconds on an IRIS 10000 (for 15 images of 30 points). The results are significantly better than those of an optimal two--image estimate. In cases when the camera projection is close to scaled orthographic, the results are comparable to those of the Tomasi/Kanade algorithm, and the algorithms are comparably fast. The algorithm incorporates a quantitative theoretical analysis of the bas--relief ambiguity and exemplifies how such an analysis can be exploited to improve reconstruction accuracy. Also, we demonstrate a structure--from--motion algorithm for partially calibrated cameras, with unknown focal length varying from image to image. Unlike the projective approac..