Studies on Free-viewpoint Image Generation of Large-scale Outdoor Environments Based on Omni-directional Dynamic Imaging

该项目的目标是使用全方位动态成像来虚拟化大规模的户外环境。本文首先构建了全方位成像系统，然后结合基于模型和基于图像的方法，提出了基于捕获图像和GPS定位的摄像机外部参数估计、场景三维重建和自由视点图像生成等算法。最后，已开发出增强虚拟的原型系统，并成功地向公众开放的实验证明。

The objective of this project is to virtualize large-scale outdoor environments using omni-directional dynamic imaging. Firstly we have constructed omni-directional imaging systems, and then developed a number of algorithms of extrinsic camera parameter estimation from captured images and GPS positioning, 3D reconstruction of scene, and free-viewpoint image generation by integrating model-based and image-based approaches. Finally prototype systems of augmented virtuality have been developed and successfully demonstrated by experiments opened to the public.