3D Reconstruction Based on Epipolar Geometry

In this paper, we propose a new method for 3D reconstruction from three cameras based on the projective geometry. If the subject is just synthesizing images from new viewpoint, 3D shape reconstruction in Euclidean space is not required, and projective reconstruction gives enough information to synthesize new viewpoint images. This means that full calibration is not needed but recovery of epipolar geometry between input cameras is sufficient. In the proposed method, three input camera images are rectified so that the vertical and horizontal directions can be completely aligned to the epipolar planes between the cameras. This rectification provides Projective Voxel Space(PVS), in which the three axis is aligned with the camera's projection direction. Such alignement simplifies the procedure for projection and back projection between the 3D space and the image planes. Taking advantage of this PVS, we apply shape-from-silhouette in the PVS to acquire bounding space of the object. The...

• Type: Conference
• Year: 2000
• Where: MVA
• Authors: Makoto Kimura, Hideo Saito
• Keywords: Camera Images, Input Camera, MVA 2000, MVA 2007, Projective Reconstruction