节点文献
基于三维重建的虚拟现实类鸟飞行系统
Virtual reality bird like flight system based on 3D reconstruction
【摘要】 随着虚拟现实(virtual reality, VR)技术的飞速发展,各种应用层出不穷。然而目前大多应用仅限于定点的静态全景展示及游览,无法利用VR化不可能为可能的优越性。本文提出了一种融合VR技术与三维重建算法的类鸟飞行交互模拟系统。该系统搭建了结构稳定的硬件控制平台用以改变实际姿态,同时通过软件仿真了虚拟环境中鸟类的飞行,结合硬件驱动及软件模拟实现了深层次的飞行模拟。现有的虚拟环境大多通过人工建模、激光扫描仪或无人机(unmanned aerial vehicle, UAV)航拍构造,其中建模存在费时费力的缺点,激光扫描仪则无法适应大规模重建,而UAV成本高昂且需专业培训。本文引入了基于图像分簇的PMVS (patch based multi-view stereopesis)算法,只需输入特定场景的图片即可利用计算机自动恢复出三维结构,不仅快速而且可对任意场景重构。使用户在足不出户的情况下即可在世界各地体验飞行,为旅游事业及VR产业提供了一种新的可能。
【Abstract】 With the rapid development of virtual reality(VR) technology, various applications emerge.However, the current applications can not make full use of the advantages of VR,and are mostly limited to fixed-point static panoramic display and tour.Therefore, this paper proposes a bird like flight interactive simulation system which integrates VR technology and 3D reconstruction algorithm.The system builds a hardware control platform with stable structure to change the actual posture, and at the same time the bird flight in the virtual environment is simulated by software, thus a deep flight simulation combined hardware driving and software simulation is realized.Most of the existing virtual environments are constructed by artificial modeling, laser scanning or unmanned aerial vehicle(UAV) aerial photography.The artificial modeling is time-consuming and laborious, the laser scanner can not adapt to large-scale reconstruction, and the UAV needs professional training.This paper introduces the patch based multi-view stereopesis(PMVS) algorithm based on image clustering, which can automatically restore the three-dimensional structure. By inputting a specific set of images, the algorithm can reconstruct any scene quickly.Users can experience to fly around the world without leaving home, which provides a new possibility for tourism and VR industry.
【Key words】 computer simulation; virtual reality(VR); structure-from-motion(SFM); multi view stereo; software and hardware development platform;
- 【文献出处】 光电子·激光 ,Journal of Optoelectronics·Laser , 编辑部邮箱 ,2023年03期
- 【分类号】TP391.9
- 【下载频次】45