Collaborative Research: CCRI: Planning: InfraStructure for Photorealistic Image and Environment Synthesis (I-SPIES)

许多计算机和信息科学与工程（CISE）研究社区利用由视觉图像或3D虚拟环境组成的数据集进行计算机视觉，机器人，多媒体系统，虚拟现实和混合现实的研究。这些数据集中的许多数据集由先前用相机和其他光学传感器捕获的图像或先前从3D虚拟环境渲染的合成图像组成。这些数据集的静态性质限制了它们的有用性和潜在应用。最近，一些研究人员提供了数据集和工具，用于使用可定制的虚拟相机位置从3D虚拟环境合成新图像，这拓宽了他们的研究应用。然而，这些数据集中的许多数据集由产生非真实感图像的较低保真度室内虚拟环境组成。此外，这些数据集缺少室外虚拟环境，目前还没有在研究界共享自定义相机位置的工具。本计划通过调查使用高质量地面激光扫描仪捕获和创建室内和室外真实世界位置的高保真逼真虚拟环境的可行性，准备解决以前数据集的局限性。这将与通过在顶级学术会议上举办的研讨会对相关CISE研究社区进行的调查相结合。该项目将使用拟议的激光扫描仪方法开发两个初步数据集，一个室内，一个室外，并确定社区需求，优先事项，以及对拟议的真实感图像和环境合成结构（I-SPIES）的支持。本科生将通过中央佛罗里达大学的EXCEL STEM项目（前NSF STEP项目）参与初步数据集的开发。该奖项反映了NSF的法定使命，并被认为值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估来支持。

Numerous Computer and Information Science and Engineering (CISE) research communities leverage datasets comprised of visual images or 3D virtual environments to conduct research in computer vision, robotics, multimedia systems, virtual reality, and mixed reality. Many of these datasets consist of either images previously captured with cameras and other optical sensors, or synthetic images previously rendered from 3D virtual environments. The static nature of these datasets limits their usefulness and potential applications. Recently, some researchers have provided datasets and tools for synthesizing new images from 3D virtual environments using customizable virtual camera positions, which broadens their research applications. However, many of these datasets consist of lower-fidelity indoor virtual environments that yield non-photorealistic images. Furthermore, such datasets are missing outdoor virtual environments, and tools for sharing custom camera positions within the research community are not currently available.This planning project prepares to address the limitations of prior datasets by investigating the feasibility of using high-quality terrestrial laser scanners to capture and create high-fidelity, photorealistic virtual environments of real-world locations, both indoor and outdoor. This will be coupled with surveys of the relevant CISE research communities through workshops held at top academic conferences. This project will result in the development of two preliminary datasets, one indoor and one outdoor, using the proposed laser-scanner methodology, and the identification of community needs, priorities, and support for the proposed InfraStructure for Photorealistic Images and Environment Synthesis (I-SPIES). Undergraduate students will be engaged in the development of the preliminary datasets through the University of Central Florida’s EXCEL STEM program, a former NSF STEP programThis award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.