3DTV View Generation Using Uncalibrated Cameras

This paper proposes a method for synthesizing free viewpoint video which is captured by uncalibrated multiple cameras. Each cameras are allowed to be zoomed and rotated freely during capture. Neither intrinsic nor extrinsic parameters of our cameras are known. Projective grid space (PGS), which is the 3D space defined by the epipolar geometry of two basis cameras, is employed for calibrating dynamic multiple cameras, because geometrical relations among cameras in PGS are obtained from 2D-2D corresponding points between views. We utilize keypoint recognition for finding corresponding points in natural scene for registering cameras to PGS. Moving object is segmented via graph cut optimization. Finally, free viewpoint video is synthesized based on the reconstructed visual hull. In the experimental results, free viewpoint video which is captured by uncalibrated cameras is successfully synthesized using the proposed method.