Instrumentation For Computer Science & Engineering Research in Image Segmentation, Registration, Modelling and Interactive Model-Based Task Optimization

马里兰州巴尔的摩约翰霍普金斯大学计算机科学与工程研究计算机科学与工程研究--图像分割、配准、建模和基于交互模型的任务优化为约翰霍普金斯大学计算机系的一个新实验室提供了计算服务器和可视化工作站，用于机器人学、图像处理、计算机视觉、建模和计算机集成外科的跨学科研究。具体的研究项目包括：(A)从3D断层图像中分割、建模和表征复杂的管状结构，例如血管和支气管树；(B)在建模的3D结构中基于视觉的准确的相机到模型的注册和导航；以及(C)开发结合数值优化、几何建模、路径规划和不确定性分析的方法，以及交互式可视化，以帮助人类专家计划复杂的3D任务以进行机器人辅助执行。这项研究将2D和3D图像和模型的数值密集计算与模型和中间计算结果的交互可视化相结合，使用生物医学建模的例子，使用生物医学成像、内窥镜和放射治疗计划中的问题来提供动机和重点。所提供的设备允许对计算机视觉、建模、计算转向和人机伙伴关系中的新算法和方法的全尺寸图像数据集和模型进行现实的实验评估，从而显著地促进了这项研究。

CDA-9529509 Taylor, Russell H. Wolff, Lawrence B. The Johns Hopkins University Baltimore, Maryland Instrumentation for Computer Science and Engineering Research in Image Segmentation, Registration, Modeling, and Interactive Model-based Task Optimization A computation server and visualization workstation are provided for a new laboratory in Computer Science Department at The Johns Hopkins University for interdisciplinary research in robotics, image-processing, computer vision, modeling, and computer-integrated surgery. Specific research projects include: (a) segmentation, modeling, and characterization of complex tubular structures such as blood vessels and bronchial trees from 3D tomographic images; (b) accurate vision-based camera-to-model registration and navigation within modeled 3D structures; and (c) development of methods for combining numerical optimization, geometric modeling, path planning and uncertainty analysis, and interactive visualizations to help a human expert plan a complex 3D task for robotically-assisted execution. This research combines numerically intensive computation on 2D and 3D images and models with interactive visualization of models and intermediate computational results, using examples from biomedical modeling, using problems in biomedical imaging, endoscopy, and radiation therapy planning to provide motivation and focus. The provided equipment significantly promotes this research by permitting realistic experimental evaluation on full-scale image data sets and models of novel algorithms and methods in computer vision, modeling, computational steering, and human-machine partnerships.