3D Reconstruction of Indoor Scenes Using RGB-D Monocular Vision

Aiming at the problem of low speed of 3D reconstruction of indoor scenes with monocular vision, the color images and depth images of indoor scenes based on ASUS Xtion monocular vision sensor were used for 3D reconstruction. The image feature extraction using the ORB feature detection algorithm, and compared the efficiency of several kinds of classic feature detection algorithm in image matching, Ransac algorithm and ICP algorithm are used to point cloud fusion. Through experiments, a fast 3D reconstruction method for indoor, simple and small-scale static environment is realized. Have good accuracy, robustness, real-time and flexibility.