Identifying the Benefits of Immersion in Virtual Reality for Volume Data Visualization
Identifying the Benefits of Immersion in Virtual Reality for Volume Data Visualization
复制标题
确定沉浸在虚拟现实中对体数据可视化的好处
DOI：
--
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
D. Bowman
中科院分区：
文献类型：
--
作者：
B. Laha;D. Bowman
研究人员传统上使用桌面显示系统来可视化和分析体积数据。部分原因是缺乏实证结果来证明沉浸式技术在体数据分析方面的优势，另外还因为高度沉浸式虚拟现实 (VR) 平台的成本。探索沉浸式好处的研究人员倾向于比较整个显示系统，而不是评估沉浸式各个组件的好处。 VR 社区需要进行受控实验来收集有关沉浸感各个组成部分的好处的经验数据。为了将结果推广到各种领域，还需要一种分类法，将使用体数据执行的任务分类为一般类别。在我们的工作中，我们开发了初步的任务分类法，并正在进行研究以确定沉浸式的各个组成部分对体数据分析任务的影响。关键词-任务分类、沉浸的好处、体积可视化、受控实验、虚拟现实、虚拟环境、沉浸式可视化。一、体积可视化和 VR 的使用 对体积格式的数据进行可视化分析和探索是各个领域研究人员的常见任务。体积可视化用于医学（例如，来自脑部扫描的功能性 MRI 数据、心脏或肺部的 CT 扫描）、细胞生物学（例如，来自共焦显微镜的数据）、地质学（例如，岩层）、古生物学（例如，化石扫描）以及许多其他学科 [1]。传统上，这些不同领域的科学家和研究人员使用台式计算机系统来可视化和分析体积数据。这些系统具有单视场渲染、较小的视场 (FOR) 和视场 (FOV)、较小的显示尺寸，并且缺乏头部跟踪渲染。许多人建议使用具有更高沉浸水平（系统 [2] 提供的感官保真度的客观水平）的虚拟现实 (VR) 系统进行科学可视化，包括体数据分析，因为沉浸式 VR 旨在以更易于理解和探索的方式显示空间复杂结构 [3]。但几乎没有经验证据证实这些说法。二.多少沉浸感才足够？为了验证沉浸式好处的说法，我们可以进行实证实验，比较不同级别的沉浸式分析体积数据可视化的有效性。探索沉浸式好处的研究人员传统上会以批发方式比较特定系统（例如，桌面、CAVE 与鱼缸 VR [4]）。这些实验对 VR 研究界具有重要价值。他们展示了沉浸式体验的好处，超越了 VR 令人印象深刻的视觉吸引力。但这些实验的结果在两个重要方面受到限制。 A. 限制 1：结果缺乏对其他 VR 系统的通用性 当整个 VR 系统相互比较时，沉浸感的几个组成部分（例如 FOR、FOV、立体视觉、头部跟踪渲染）在不同条件下同时变化。如果这样的研究确定了沉浸式的好处，我们就无法知道沉浸式的哪些组成部分或组成部分的组合带来了这些好处。由于这种混淆，我们无法将结果推广到所研究的特定系统之外的 VR 系统。我们不知道具有中等沉浸感的 VR 系统是否可以提供与高度沉浸式系统相同的好处。这些细微差别的重要性源于高度沉浸式 VR 硬件（例如 CAVE 或头戴式显示器 (HMD)）的昂贵成本。此外，考虑到更便宜的商用 VR 硬件可以提供中等程度的沉浸感，我们需要更细粒度的实证结果来确定此类系统是否可能具有更有利的成本效益比。 B. 限制 2：结果缺乏对其他领域的通用性第二个限制是由于实验必须使用来自特定领域的数据集和任务来评估沉浸式的好处。因此，很难将结果应用到其他领域和任务。例如，一项研究表明沉浸式 VR 有利于分析体积脑部扫描数据，但对地质学家来说意义不大。为了让更广泛的受众认识到沉浸式虚拟现实对体积数据分析的好处，我们需要以一种可推广到各个学科的方式来确定沉浸式的好处，但通过评估所有可能的数据集、任务和领域来做到这一点是不切实际的。如果我们要对沉浸式的好处做出更普遍的主张，我们需要更深入地了解体数据集可视化分析所涉及的任务。
Researchers have traditionally used desktop display systems for visualizing and analyzing volume data. This is partially due to the lack of empirical results showing benefits of immersion for analysis of volume data, and also due to the cost of highly immersive virtual reality (VR) platforms. Researchers exploring the benefits of immersion tend to compare entire display systems rather than evaluating the benefits of individual components of immersion. The VR community needs controlled experimentation to gather empirical data on the benefits of individual components of immersion. In order to generalize the results to a variety of domains, a taxonomy that classifies tasks performed with volume data into general categories is also needed. In our work, we have developed a preliminary task taxonomy and are performing studies to identify the effects of various components of immersion on volume data analysis tasks. Keywords-Task taxonomy, benefits of immersion, volume visualization, controlled experiments, virtual reality, virtual environments, immersive visualization. I. VOLUME VISUALIZATION AND THE USE OF VR Visually analyzing and exploring data in volumetric format is a common task for researchers from various domains. Volume visualization is used in medicine (e.g., functional MRI data from brain scans, CT scans of the heart or lungs), in cell biology (e.g., data from confocal microscopy), in geology (e.g., rock strata), in paleontology (e.g., fossil scans), and in many other disciplines [1]. Traditionally, scientists and researchers in these various domains have used desktop computer systems for visualizing and analyzing volume data. These systems have monoscopic rendering, a small field of regard (FOR) and field of view (FOV), and a small display size, and lack head-tracked rendering. Many people have suggested using virtual reality (VR) systems with higher levels of immersion (the objective level of sensory fidelity provided by a system [2]) for scientific visualization, including the analysis of volume data, since immersive VR is designed to display spatially complex structures in a manner easier to understand and explore [3]. But there is little empirical evidence validating these claims. II. HOW MUCH IMMERSION IS ENOUGH? To validate the claims of the benefits of immersion, we can run empirical experiments comparing the effectiveness of different levels of immersion for analyzing visualizations of volume data. Researchers exploring benefits of immersion have traditionally compared specific systems in a wholesale fashion (e.g., desktop vs. CAVE vs. fishtank VR [4]). These experiments are of great value to the VR research community. They demonstrate the benefits of immersion beyond the impressive visual appeal of VR. But the results of these experiments are limited in two important ways. A. Limitation 1: Lack of generalizability of results to other VR systems When entire VR systems are compared to one another, several components of immersion (e.g., FOR, FOV, stereoscopy, head-tracked rendering) vary simultaneously between conditions. If such a study identifies a benefit of immersion, we cannot know which component(s) or combination of components of immersion resulted in those benefits. As a result of this confound, we cannot generalize the results to VR systems beyond the specific systems that were studied. We do not know whether VR systems with an intermediate level of immersion might have delivered the same benefits as a highly immersive system. The importance of these finer differentiations stems from the costliness of highly immersive VR hardware such as CAVEs or headmounted displays (HMDs). Also, given the availability of cheaper commodity VR hardware offering moderate levels of some components of immersion, we need finer-grained empirical results to determine whether such systems might have a more favorable cost-benefit ratio. B. Limitation 2: Lack of generalizability of results to other domains The second limitation arises due to the fact that experiments must evaluate the benefits of immersion using datasets and tasks from specific domains. Thus, it is difficult to apply the results to other domains and tasks. For example, a study showing that immersive VR is beneficial for analyzing volumetric brain scan data is of little import to the geologist. To realize the benefits of immersive virtual reality for volumetric data analysis for a broader audience, we need to establish the benefits of immersion in a manner generalizable across various disciplines, but it is impractical to do this by evaluating all possible datasets, tasks, and domains. We need a deeper understanding of the tasks involved in visual analysis of volume datasets if we are to make more general claims about the benefits of immersion.