Bullet-time image generation without 3-D

We propose a new method for generating bullet-time images without any 3-D information. In a general way, we estimate 3-D camera pose and a 3-D position of a target point, then compute a homography that moves the target point to the image center. In this work, we re-parameterize the homography without 3-D information of the camera and the target point. We also estimate the homography by using an image matching technique. We compare the proposed method with an existing method that uses 3-D information and show the effectiveness of our method.