3D Object Reconstruction with Kinect Based on QR Code Calibration

摘要

We propose a method of 3D reconstruction of small-sized object based on Kinect V2 RGB-D camera and turntable, which eliminates the need of costly feature extraction and robust matching techniques for motion estimation. Identification and detection of a QR code are used to calibrate the system, and on this basis, point cloud coordinate conversion and background removal are realized. Our coarse registration algorithm uses the fixed rotation angle of the turntable to construct the rotation matrix between frames. Combined with ICP (Iterative Closest Point) algorithm for precise registration, the object point cloud model is obtained. We achieve a cost-effective, convenient and practical 3D reconstruction process for small-sized objects. Experimental results show that the method can stably and effectively obtain 3D models of objects that are small and difficult to extract features, which has certain application value in product display.