Motion detection from a moving observer has been a very important technique for 3-D dynamic image analysis, especially in the research of obstacle detection and tracking for autonomous driving systems and driver supporting systems. Because of the continuously changing background, detecting the real moving objects has become very difficult and always employs optical flow methods to measure the flow vectors difference between background and the moving objects. Unfortunately, there is a huge calculation cost to obtain accurate optical flow vectors. A new motion detecting method based on camera 3-D motion analysis is proposed in this paper. With the special motion feature of the on-board camera and its Focus of Expansion (FOE), our method can detect the real moving objects from the continuously changing background scene by simply using the pure feature matching between two adjacent frames. The camera 3-D motion can theoretically be determined in our method by using only three pairs of matching points, which make it faster and more efficient for real-time applications. Experiments on real outdoor road scenes show the accuracy and efficiency of our approach.
展开▼
