A new multistage approach to motion and structure estimation by gradually enforcing geometric constraints

摘要

The standard 2-stage algorithm first estimates the 9 essential parameters defined up to a scale factor and then refines the motion estimation based on some statistically optimal crierial. We propose in this paper a novel approach by introducing an intermediate stage which consists in estimating a 3 x 3 matrix de-fined up to a scale factor by imposing the rank-2 constraint (the matrix has seven indepent parameters). The idea is to gradually project parameters estimated in a high dimensional space onto a slightly lower-dimensional space, namely from 8 dimensions to 7 and finally to 5.