Edge-based Adaptive Real-Time 3D Tracking

3D tracking is very important in a lot of fields like visual servoing of robotic arms and Augmented Reality. Among all methods of 3D tracking, computer vision offers solutions that are non-invasive, accurate, and low-cost, which attract a lot of research. We briefly dicuss different solutions of vision based 3D tracking, and introduce edge-based tracking in detail. Then we talk about the innovations of this paper and show some experimental results.We propose an edge-based adaptive real-time 3D tracking method, which can effectively handle the tracking of textureless objects. With a prior 3D model of the tracking object, the proposed method can effectively detect and track the object edges using temporal motion information, and then accurately estimate the extrinsic camera parameters. Our contributions are summarized as follows:1)We utilize temporal information for motion prediction and adaptive edge detection, which can significantly improve the robustness of edge tracking for fast object movement; 2)We propose a RANSAC-based edge matching strategy, which can effectively eliminate outliers to make the tracking of complex objects robust;3)By extracting the silhouette from the prior 3D model, we extend the edge-based tracking from CAD model to general mesh model. Experimental results demonstrate the robustness and efficiency of our method, which can satisfy the demand of augmented reality and virtual assembly.