Computer Graphics Tools for Understanding Tensor-Valued Volume Data: A Painting Metaphor

这个多学科的研究项目将调查、应用和评估新的计算机图形工具，以理解向量和二阶张量值的体积数据。这些工具的有效性将通过它们对几个特定物理现象的洞察程度来评估：充分发展的湍流，多发性硬化症小鼠模拟的疾病进展，以及小鼠大脑的新生发育。新技术的应用以及与其他领域研究人员的合作为验证这些技术并确保它们在解决科学问题方面作出反应提供了机会。这些技术将使用绘画比喻来开发。几个世纪以来，艺术家们一直使用分而治之的方法来创作绘画，用不同的画层来表现场景的不同方面。此外，艺术家使用笔触在画布上表现场景的重要特征，描绘场景的本质并消除不必要的细节，这些笔触具有纹理、大小、形状、颜色和不透明度等许多特征。这两种技术以及在使用它们时的集体艺术体验都可以应用于显示体积数据的计算机图形学。理解矢量值和张量值的体积数据的技术尤其有益，因为体积包含如此多的信息，而且张量值和矢量值有许多相关、相互依赖的方面。现有技术在有效提取数据和同时显示数据的许多相关方面的能力有限，但新技术将解决这两个问题。因此，他们将更有效地产生对测量的物理现象的洞察，以及在交流洞察方面。由于这些技术将在两个学科中应用于三个具体问题，因此它们可能会有更广泛的应用。

This multi-disciplinary research project will investigate, apply, and evaluate new computer graphics tools for understanding vector- and second-order tensor-valued volume data. The efficacy of the tools will be evaluated by the extent to which they generate insight into several specific physical phenomena: fully-developed turbulent flow, the progression of disease in a mouse analogue of multiple sclerosis, and the neonatal development of the mouse brain. The application of the new techniques and collaboration with researchers in other fields provides the opportunity to validate the techniques and ensure that they are responsive in addressing the scientific problems. The techniques will be developed using a painting metaphor. For centuries artists have used a divide-and-conquer approach to creating paintings, representing different aspects of a scene in different layers of paint. In addition, artists use brush strokes, with their many characteristics of texture, size, shape, color, and opacity, to represent important features of a scene on a canvas, depicting the essence of the scene and eliminating unnecessary detail. Both techniques, and the collective artistic experience in using them, can be applied to computer graphics in the context of displaying volume data. Techniques for understanding vector-valued and tensor-valued volume data stand to benefit particularly, because volumes contain so much information and because tensor and vector values have many relevant, interdependent aspects. Existing techniques are limited in their ability to abstract the data effectively and to display the many relevant aspects of the data simultaneously, but the new techniques will address both of these problems. As a result, they will be more effective in generating insight into the physical phenomena measured as well as in communicating the insight. Because the techniques will be developed with application to three specific problems in two disciplines, they are likely t o have broader application.