An Efficient Trinocular Rectification Method for Stereo Vision

In this paper, we introduce an efficient method of trinocular rectification to enhance the performance of stereo vision. In order to perform the trinocular rectification, we first obtain trinocular images by three cameras lying on a plane that faces the target objects. Then, we extract the original camera parameters, and compute the new camera parameters that transform the epipole on each image to the point at infinity. After that, we determine the homographies which relate the original camera parameters to the new ones, and decide an affine matrix which ensure 45 degrees for the epipolar line between top and left image. Finally, the trinocular rectification is finished by applying homographies and an affine transform to the obtained trinocular images. In contrast to conventional trinocular rectification methods that are mathematically and geometrically complicated, the proposed method is simple but efficient and accurate.