Real-Time 3D Reconstruction Using a Kinect Sensor

摘要

Nowadays, in the robotics industry, the point cloud processing gets more and more well-known due to the launching of Microsoft Kinect device in 2011. This paper aims to compare the methods for 3D reconstruction using a Kinect sensor and to rebuild a detailed model from an indoor scene in order to use it in a CAD application. The acquisition system allows the user to rotate around the target area and see a continuously updated 3D model of the desired object. For creating the final model, different viewpoints must be acquired and fused together in a single representation.