Multiple video camera calibration using EPI for city modeling
Hiroshi Kawasaki · Medical Entomology and Zoology · 2004
In this paper, we present an Epipolar Plane Image (EPI)based multiple video camera calibration method which enables simultaneous estimation of the multiple video cameras’ parameters and the synchronization of video data. Using our proposed method, in order to capture a large scale scene’s texture image, a user is only required to install multiple video cameras on top of his/her car without setting exact configurations, and can take a video without using a special external synchronization device. Since the optical centers of cameras are usually inside the cameras, and bringing them together is impossible, distortions inevitably occur in fused images. However, with our proposed method, the optical centers of the multiple cameras are made to coincide at one point in spatio-temporal space and thus we can fuse these images into a single one with no distortions. Another strength of our method is that we do not need any overlapping areas between images for calibration, therefore large scale scene can be efficiently captured.