3D reconstruction from bifocus imaging

This paper presents a method to reconstruct the three-dimensional model in monocular stereo vision. Depth can be obtained from monocular disparity in bifocus images. The stereo disparity results from lenses´ separation during imaging. Besides, 3D information can be acquired with depth and camera intrinsic parameters by means of camera calibration. Based on the characteristics of bifocus imaging, this paper develops a new method to detect and match feature, which fits monocular stereo vision much better. With only a pair of bifocus images and a calibrated camera, three-dimensional model of scene is constructed successfully.