Combination of edge element and optical flow estimates for 3D-model-based vehicle tracking in traffic image sequences

A model-based vehicle tracking system for the evaluation of inner-city traf fic video sequences has been systematically tested on about 15 minutes of r eal world video data. Methodological improvements during preparatory test p hases affected-among other changes-the combination of edge element and opti cal flow estimates in the measurement process and a more consequent exploit ation of background knowledge. The explication of this knowledge in the for m of models facilitates the evaluation of video data for different scenes b y exchanging the scene-dependent models. An extensive series of experiments with a large test sample demonstrates that the current version of our syst em appears to have reached a relative optimum: further interactive tuning o f tracking parameters does no longer promise to improve the overall system performance significantly. Even the incorporation of further knowledge rega rding vehicle and scene geometry or illumination has to cope with an increa sing level of interaction between different knowledge sources and system pa rameters. Our results indicate that model-based tracking of rigid objects i n monocular image sequences may have to be reappraised more thoroughly than anticipated during the recent past.