We want to use a large-scale camera array system in which each camera is placed at the desired position to photograph a subject and later render images of the subject viewed from various directions or render images for a three-dimensional display. The homography matrix for each camera should be calculated in advance to correct the captured images. In the case that each camera is physically facing toward the subject as precisely as possible but the captured image still includes geometrical distortion, if the expected error in the deviations from the ideal directions is assumed to be the zero vector, the homography matrix of each camera can be easily obtained.
展开▼
