ESTIMATION OF DISPLACEMENTS FROM 2 3-D FRAMES OBTAINED FROM STEREO

We present a method for estimating 3-D displacements from two stereo frames. It is based on the hypothesize-and-verify paradigm used to match 3-D line segments between the two frames. In order to reduce the complexity of the method, we make the assumption that objects are rigid. We formulate a set of complete rigidity constraints for 3-D line segments and integrate the uncertainty of measurements in this formulation. The hypothesize-and-verify stages of the method use an extended Kalman filter to produce estimates of the displacements and of their uncertainty. In the experimental sections, the algorithm is shown to work on indoor and natural scenes. Furthermore, it is easily extended, as is also shown, to the case where several mobile objects are present. The method is quite robust, fast, and has been thoroughly tested on hundreds of real stereo frames.