CISE Research Instrumentation: Video Analysis and Reconstruction in a Distributed Environment

9818332Jaynes, ChristopherSeales， BrentUniversity of kentucyuniversity分布式环境中的视频分析和重建该研究仪器支持以下研究项目：分布式异构网络中的视频分析和传感器控制，自动站点模型获取和重建，实时合成环境可视化和用户引导错误纠正，以及非结构化可视化环境中的实时深度扭曲。该合同支持肯塔基大学计算机科学系正在进行的研究，并购买三个组件：主动控制立体视觉传感器平台、高性能图像处理计算服务器和实时可视化环境。视频分析和重建项目旨在开发分布式实时方法来解决基于大规模视频和图像的问题。该项目利用高速计算机网络连接的三个组件平台：主动传感器平台、分布式高性能计算环境和灵活的实时显示系统。实时感知、重建和可视化环境的总体目标导致了几个需要使用一个或多个平台的特定问题。主动传感器平台允许研究传感器控制和大规模图像数据压缩和通信中的问题。通常，数据集可能包含数千个复杂环境的图像。一个高性能的计算平台，结合复杂的计算机视觉算法，是实时重建这些环境所必需的。柔性显示系统使合成环境的可视化研究成为可能。复杂的用户与模型的交互可以通过将用户指示的错误从可视化平台传递给传感器平台以获取新数据，或传递给计算服务器以根据新信息对数据进行重新处理，从而改进模型。这种平台之间的实时交互是通过高速网络实现的。

9818332Jaynes, ChristopherSeales, BrentUniversity of KentuckyVideo Analysis and Reconstruction in a Distributed EnvironmentThis research instrumentation enables research projects in:- Video Analysis and Sensor Control in Distributed Heterogeneous Networks,- Automatic Site Model Acquisition and Reconstruction,- Real-Time Synthetic Environment Visualization and User-Guided Error Correction, and- Real-Time Depth Warping for Unstructured Visualization Environments.This award supports on-going research in the Department of Computer Science at the University of Kentucky and contributes to the purchase of three components: Active-Control Stereo Vision Sensor Platform, High Performance Image Processing Compute Server, and a Real-Time Visualization Environment. The video analysis and reconstruction project seeks to develop distributed real-time approaches to large-scale video and image based problems. The project utilizes three component platforms connected by a high-speed computer network: an active sensor platform, a distributed high performance computing environment, and a flexible real-time display system. The overall goal of sensing, reconstructing, and visualizing an environment in real-time leads to several specific problems that require the use of one or more of the platforms. The active sensor platform allows investigation of sensor control and issues in the compression and communication of large-scale image data. Typically, datasets may contain thousands of images of a complex environment. A high-performance computing platform, in conjunction with sophisticated computer vision algorithms, is necessary for the reconstruction of these environments in real-time. The flexibledisplay system enables research in the visualization of synthetic environments. Sophisticated user interaction with the model can lead to model improvement by the communication of user-indicated errors from the visualization platform to the sensor platform for the acquisition of new data, or to the compute server, for reprocessing of the data in light of the new information. This real-time interaction between platforms is facilitated via the high-speed network.