Recovery of Viewer-Centered and Object-Centered Depth from 2-D Projections

从二维投影恢复以观察者为中心和以对象为中心的深度

• 批准号: 8819565
• 负责人: Myron Braunstein
• 金额: $ 16.36万
• 依托单位: University of California-Irvine
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1989
• 资助国家: 美国
• 起止时间: 1989-02-15 至 1992-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8819565&HistoricalAwards=false
• 关键词: Recovery; Viewer; Centered; Object; Depth

The visual environment is three-dimensional, but our information about it is based on two-dimensional retinal projections. This research will investigate two types of processes for recovering three-dimensional relationships from two-dimensional images. The first type of process recovers distances from the eye to features in the environment. The research will examine two examples of this type of process: stereoscopic vision and motion parallax (the perception that objects moving faster in the retinal projection are closer to the eye). The second type of process recovers three-dimensional relationships among objects in the environment independently of distances from the eye. "Structure- from-motion" is the example of this type of process that will be considered. Perceptions based on structure-from-motion recover three-dimensional relationships among objects directly from the relative motions of the objects, assuming only that the objects are moving rigidly or at constant speeds. Information about three-dimensional distances from the eye and information about three-dimensional relationships among objects must be integrated to provide a complete perception of the three-dimensional environment. Preliminary research has shown that structure- from-motion facilitates the use of stereoscopic information. A series of five experiments will examine the basis for this facilitation. Preliminary observations have also indicated that similar relative motions projected at the eye may lead to different three-dimensional perceptions, depending on whether a motion parallax analysis (projected velocities based on distance from the eye) or a structure-from-motion analysis (velocities based only on relationships among objects) is applied by the observer. Three experiments will examine the conditions that determine which of these alternative interpretations is applied by human observers. An understanding of how different types of information for perceiving three-dimensional relationships from two-dimensional images are combined has relevance to such questions as: Can people lacking stereoscopic vision when tested with static displays demonstrate stereoscopic ability when motion is added to the displays, and how should the three-dimensional environment be represented in artificial displays requiring accurate observer responses, such as flight simulator displays and aircraft displays that represent the virtual environment?

视觉环境是三维的，但我们关于它的信息是基于二维视网膜投影的。本研究将研究从二维图像中恢复三维关系的两种过程。第一种类型的过程恢复从眼睛到环境中特征的距离。这项研究将考察这类过程的两个例子：立体视觉和运动视差(物体在视网膜投影中移动更快的感觉离眼睛更近)。第二种类型的处理独立于距眼睛的距离来恢复环境中对象之间的三维关系。“从运动到结构”就是将被考虑的这种类型的过程的例子。基于运动结构的感知直接从对象的相对运动中恢复对象之间的三维关系，仅假设对象以刚性或恒定速度移动。关于距眼睛的三维距离的信息和关于物体之间的三维关系的信息必须被整合，以提供对三维环境的完整感知。初步研究表明，运动结构有助于立体信息的使用。一系列的五个实验将检验这种促进作用的基础。初步观察还表明，投射在眼睛上的类似相对运动可能会导致不同的三维感知，这取决于观察者是否应用了运动视差分析(基于眼睛距离的投影速度)或运动结构分析(仅基于对象之间的关系的速度)。三个实验将检验决定人类观察者应用这些可选解释中的哪一个的条件。理解如何组合用于从二维图像感知三维关系的不同类型的信息与这样的问题相关：当用静态显示器测试时缺乏立体视觉的人在向显示器添加运动时是否表现出立体能力，以及三维环境应该如何在需要准确的观察者响应的人造显示器中表示，例如代表虚拟环境的飞行模拟器显示器和飞机显示器？