CAREER: High Quality and Efficient Rendering of Discrete Primitives for Interactive Visualization

在过去的几十年里，大多数以前的图形和可视化系统都是基于多边形的。然而，随着科学模拟和先进的生物医学和工业采集设备的出现，越来越多的空前大小的物体以离散样本作为自然表征。在本提案中，PI将推广图像重采样理论来描述图形绘制的基本重采样问题。在这一理论的指导下，将设计实用的方法来执行高质量和高效的重采样，并进一步应用于各种离散对象的渲染，包括二维图像扭曲、体渲染和基于点的渲染。发展的理论也将用于指导新的基于点的表示的设计，使离散的原语和多边形可以在一个统一的呈现系统中呈现。新设计的算法将通过应用程序进行验证和传播，主要应用于PI已经合作的两个实际应用程序。体绘制算法将用于可视化大型模拟，如空气动力学、天体物理学和化学污染扩散等领域。基于点的建模和绘制算法将用于从现代激光距离扫描中构建基于点的模型，并最终为数字建筑设计提供一个交互式系统。这种验证和传播是通过实施一项强有力的教育计划来实现的，该计划采取了一些互补的举措：(1)在明尼苏达大学开设新的研究生和本科课程；（二）开发教学软件；(3)其他活动，包括组织研讨会和辅导课。智力优势：使用重采样理论连贯设计的算法通过提供更高的质量和更好的效率，代表了科学可视化系统的改进的渲染技术。提出的研究很好地建立在PI在图像和体积重采样和抗锯齿以及加速体积和多边形渲染方面的扎实知识和丰富经验之上。更广泛的影响：所提出的职业发展计划不仅将推动科学可视化和计算机图形学的最新发展，而且除了在本项目过程中用于方法验证和传播之外，还将对许多其他由先进科学可视化促进的研究领域做出重大贡献。这些领域的可能性可能会开放，刺激比最初设想的更多的技术应用。

Over the past decades, most of the previous graphics and visualization systems are polygon based. However, with the advent of scientific simulations and advanced biomedical and industrial acquisition devices, more and more objects of unprecedented size are produced with discrete samples as natural representations. In this proposal, the PI will generalize image resampling theory to describe the fundamental resampling problem of graphics rendering. Guided by this theory, practical approaches to performing high quality and efficient resampling will be designed and further applied to the rendering of various discrete objects, including 2D image warping, volume rendering, and point-based rendering. The developed theory will also used to guide the design of new point-based representation, so that discrete primitives as well as polygons can be rendered in a unified rendering system. The newly designed algorithms will be validated and disseminated through application to mainly two real-world applications that the PI has already been collaborating on. The volume rendering algorithms will be used to visualize large simulations, in areas such as aerodynamics, astrophysics, and chemical contamination dispersion. The point-based modeling and rendering algorithms will be used to build point-based models from modern laser range scans and ultimately provide an interactive system for digital architecture design. This validation and dissemination are achieved by pursuing a vigorous education plan, undertaking a number of complementary initiatives: (1) developing new graduate and undergraduate courses at the University of Minnesota; (2) developing educational software for teaching; and (3) other activities including organizing workshops and tutorials. Intellectual Merits: The algorithms coherently designed using the resampling theory represent improved rendering techniques for scientific visualization systems by delivering higher quality and better efficiency. The proposed research is well built upon the PI's solid knowledge and extensive experience with image and volume resampling and antialiasing, as well as accelerated volume and polygon rendering. Broader Impacts: The proposed career development plan will not only advance the state-of-the-art of scientific visualization and computer graphics, but also make significant contributions to many other research fields which are facilitated by advanced scientific visualization, besides the ones being used for methodology validation and dissemination during the course of this project. Possibilities in these fields will likely open, stimulating even more applications of the technology than originally conceived.