Detection of Independent Motion Using Directional Motion Estimation

It is shown that the problem of independent motion detection can be addressed by analyzing constraints on low-dimensional directional (projected) components of ow elds. We construct a robust algorithm, implemented as a recursive lter, to extract directional motion parameters from long image sequences. Based on this a qualitative approach is described to detect independent motion, involving a combination of robust line-tting and one-dimensional search. The low-dimensional projections onto subspaces facilitate eecient dynamic self-adaptation of detection thresholds to achieve good performance under changing operational conditions. The analysis is extended to long image sequences by incorporating tracking and spatio-temporal ltering. The approach is applicable to general camera motion and cluttered scenes using a wide range of camera elds of view. It does not require point correspondences and has a simple extendsion to stereo. We demonstrate it on a variety of real image sequences.