3D reconstruction using Kinect v2 camera

Abstract

Kinect is an easy to use and a ordable RGB-D acquisition device that provides both spatial and color information for captured pixels. That makes it an attractive alternative to regular 3D scanning devices that usually cost signi cantly more and do not provide color info. Second generation of Kinect (v2) provides even better quality depth and color images to user. This thesis describes and implements method for 3D reconstruction using Kinect v2. Method suitability for various objects is tested and analyzed. In most circumstances the method provided satisfactory reconstructions unless very high resolution is desired. However some limitation were observed. Reflective and transparent surfaces cause failure due to depth capturing technology in Kinect v2, symmetric objects cause problems for described frame registration algorithm. For better understanding, Kinect v2 depth measuring process is described.