CISE Research Instrumentation: A High-Performance Computing and Imaging Facility for Research in Image Understanding and Analysis

9818212Boyer, Kim L.Flynn, Patrick j .俄亥俄州立大学cise研究仪器：用于图像理解和分析研究的高性能计算和成像设备该研究仪器支持以下研究项目：-机器学习的感知组织和图像注释-高级对象识别：自由形式对象和大型模型库，以及-智能图像压缩。为了支持上述项目，该奖项将用于购买高性能计算、数据采集和显示设备：几台奔腾、交换机、磁盘、备份磁带驱动器、图像分析工作站、扫描仪、数字视频磁盘、数码相机、打印机、显示器等。这项研究是俄亥俄州立大学电子工程系和测绘中心的合作成果。该设备可用于图像和视频理解和分析的研究，特别是所引用的研究项目。由于涉及大量数据集和复杂的算法，计算机视觉和图像分析的工作是高度计算密集型的。所有这些计算活动都夹在数据采集（成像）步骤和结果的显示或其他传播之间。该设施将为基于贝叶斯网络形式的图像和视频理解分布式算法的开发和实现，以及高质量数字图像和视频的获取提供一个灵活而强大的测试平台。为此，该设施包括一台商用超级计算机（Beowulf集群）和具有足够存储和网络带宽的图像分析工作站，以处理图像和视频处理数据集和算法的需求。由于所研究的许多应用涉及高分辨率航空图像，（近）实时需求或运动，因此这种能力和灵活性至关重要。为了最大限度地发挥安装的效用，包括许多输入源和图像存储设施，如数码相机和摄像机、框架存储、卫星数字视频接收器和3D扫描仪。

9818212Boyer, Kim L.Flynn, Patrick J.The Ohio State UniversityCISE Research Instrumentation: A High-Performance Computing and Imaging Facility forResearch in Image Understanding and AnalysisThis research instrumentation enables research projects in:- Perceptual Organization and Image Annotation with Machine Learning- Advanced Object Recognition: Free-Form Objects and Large Modelbases, and- Intelligent Image Compression.To support the aforementioned projects, this award contributes to the purchase of high performance computing, data acquisition, and display facilities: several Pentiums, switches, disks, backup tape drives, image analysis workstations, scanners, digital video disk, digital cameras, printers, monitors, etc. The research constitutes a cooperative effort of the Department of Electrical Engineering and the Center for Mapping at Ohio State University. The equipment enables studies in image and video understanding andanalysis, specifically the research projects cited. Work in computer vision and image analysis is highly computationally intensive, owing to the large datasets and complex algorithms involved. All of this computing activity is sandwiched between the data acquisition (imaging) step and the display or other dissemination of results. The facility will provide a flexible and powerful testbed for the developmentand implementation of distributed algorithms for image and video understanding, based on Bayesian network formalisms, as well as the acquisition of high-quality digital imagery and video. To this end, the facility incorporates a commodity supercomputer (Beowulf cluster) and image analysis workstations with sufficient storage and network bandwidth to handle the demands of image and video processing datasets and algorithms.Because many of the applications studied involve high resolution aerial imagery, (near) real-time demands, or motion, this capability and flexibility is crucial. To maximize the utility of the installation, many input sources and image storage facilities, such as digital cameras and camcorders, framestores, satellite digital video receiver, and 3D scanner are included.