Depth-color image registration for 3D surface texture construction using kinect camera system

摘要

Today some camera systems provide various opportunities to the scientists in computer vision since they capture color and depth images of a scene simultaneously. This paper presents a new 3D model construction and surface texture mapping technique for real object images captured by Microsoft (MS) Kinect camera system. Our ultimate goal is to construct textured 3D model of the real objects. To achieve this goal, we perform depth-color image registration based on “Scale Invariant Feature Transform” (SIFT) and “Speeded-Up Robust Features” (SURF) features, Hough transform, and least squares optimization. After the registration, using the depth image, non-textured 3D model is created and finally color image is mapped on top of the 3D model of the object based on Delaunay triangulation. In the experiment section, three objects having different sizes and shapes are examined. Their textured 3D models are constructed without significant problems.