Adaptive methods of moving car detection in monocular image sequences

Computer vision applications for traffic scene analysis and autonomous navigation (driver support) require highly sophisticated sensors and computation methods – they constitute a real challenge for image analysis systems. Common to both applications is the moving object detection/tracking task. In this paper we study this task on four different data abstraction levels: image segmentation, 2-D object tracking, model-based 3-D object tracking and many-object traffic scene description. Two meanings of the term ”adaptive” are considered: learning algorithms or connectionist systems and recursive estimation for dynamic systems. Generally the first approach may be applied for lowand segmentation-level analysis of finite image sequences, whereas the second approach is suitable for 2-D and 3-D object tracking and estimation.