3D RECONSTRUCTION FROM MULTIPLE IMAGES USING SINGLE MOVING CAMERA
MetadataShow full item record
A new technique to reconstruct an object or environment in 3D using a single moving camera was developed. The technique requires only one camera moving through the environment. Once a set of images or a video is captured, a point cloud and the camera path of the environment or the object can be generated. The camera has to be calibrated in advance. The procedure of 3D reconstruction consists of the SIFT feature matching algorithm, the Epipolar geometry reconstruction, fundamental matrix algebra, points cloud triangulation, and the estimation of the camera motion. The algorithm explored in this thesis is only based on visual information captured by a single moving RGB-camera.