3D scene analysis by real-time stereovision

A new solution is proposed for 3D scene analysis in security and surveillance applications. It is based on binocular stereo-vision using a prediction-verification paradigm. Adaptive change-motion detection is performed at video rate to detect moving objects in the scene. 3D information is recovered by "scanning" the scene along parallel planes at different heights. Prediction of stereo correspondence is performed through homographic transformation on the set of the selected 3D planes, to verify whether the detected change is really a moving 3D object crossing that plane, or it is just a phantom caused by shadows or highlights. Object segmentation and tracking is performed on the projected ground floor by clustering the detected features at different heights.