Optical Flow-Based Moving-Static Separation in Driving Assistance Systems

This paper introduces a novel approach for a moving static separation, which can be efficiently used for outlier removal within optical flow-based 3D reconstruction or moving object detection even in the case of a moving observer. The contributions of the proposed approach include an efficient implementation, high accuracy and embedded friendliness. These characteristics make the approach very attractive for automotive safety applications based on camera systems. The paper also highlights some possible applications where the use of the moving static separation can play a key role on improving the performance.