Estimation of depth and 3D motion parameter of moving object with multiple stereo images

Abstract

In this paper a use of stereo motion sequences is considered to estimate both three-dimensional (3D) motion and depth of 3D moving points. The problem is formulated as optimization of a cost function, accumulated sum of squared differences (SSD) computed from stereo motion sequences. It is assumed that the 3D motion is purely translational within any image window in which the SSD is computed. This method does not make any assumptions about stereo correspondences or spatial segmentation into a rigid body. We derive the unique condition under which 3D motion and depth ambiguities can be resolved. Once the ambiguities have been resolved, the 3D motion parameters and depth are estimated by a least squares technique. By analyzing the statistical characteristics of the cost function, it is shown that the precision of the estimates can be improved from data redundancy. A recursive algorithm is presented to reduce memory overload and to sequentially update the estimates. Simulation with synthetic images and experimentation with a real image sequence are presented to show the effectiveness of this method.