3D Scene Reconstruction Based on Uncalibrated Image Sequences

3D scene reconstruction is an important technique in the computer vision field. Our system can give the user a platform to reconstruct a 3D model of scene from a set of uncalibrated images which are gained by a commonly used camera. There are many key techniques in 3D reconstruction from uncalibrated image sequences, including feature matching, fundamental matrix estimation, projective reconstruction, camera self-calibration, dense stereo matching and Euclidean reconstruction. The paper is focused on above associated issues and improved some key algorithms. The effectiveness of our algorithms is evaluated in the experiments with many real image sequences.