A novel algorithm for retrieving complete 3D structure from freely captured image sequences

摘要

A novel algorithm to retrieve the complete structure from long image sequences captured by a hand held camera is proposed. Firstly, images taken around the target are divided into several subsets. Each subset has common feature points. Secondly, Euclidean reconstruction is obtained by iterative factorization with all of these points visible in each image of a certain subset. Then results coming from different subsets are brought into a common coordinate frame by similarity transformations. Finally, global optimization is applied to refine the data and produce a jointly optimal 3D structure. A significant merit of the algorithm is that it sets no restrictions on the input image sequence. The algorithm has been tested on both simulated data and real image sequences, and very realistic 3D models are obtained.