The Analysis of Optic Flow Without Locally Smooth Velocity

当只有一个物体存在时，目前大多数关于人们如何感知运动物体的3D形状的模型都能够确定物体的形状。然而，在现实世界中，我们经常会遇到许多物体在移动的场景。这项研究将开发一种新的模型，当几个物体在场景中移动时，物体的形状感知。这项研究将研究在静态视角下无法区分重叠对象的场景，但在观察者或对象运动时可以区分重叠对象的场景。三组实验将确定人类观察者在感知多个运动物体方面的局限性。第一组实验将研究人们确定在复杂场景中移动的物体数量的能力。第二组实验将研究人们如何识别场景中运动物体的形状。第三组实验将研究人们如何判断场景中物体之间的深度大小。这些结果将有助于模型的三阶段发展。第一阶段将使用第一组实验的结果，从数学上推导出在计算机生成的场景中移动的独立对象的数量。第二阶段将使用第二组实验的结果来数学描述每个物体的小区域的形状。第三阶段将使用第三组实验的结果来指定场景中对象之间距离的量化值。了解视觉系统如何确定多个运动物体的形状和深度将对用于培训飞行员的计算机生成显示器的设计、盲人视觉辅助工具的开发以及机器人视觉系统的开发具有重要意义。

Most current models of how people perceive the 3-D shape of moving objects are capable of determining shape when only a single object is present. However, in the real world we usually encounter scenes in which many objects are moving. This research will develop a new model of the perception of the shape of objects when several objects are moving in a scene. The research will study scenes in which overlapping objects cannot be distinguished in a static view, but can be distinguished when the observer or the objects are in motion. Three sets of experiments will determine the limitations of human observers in perceiving multiple moving objects. The first set of experiments will study people's ability to determine the number of objects that are moving in a complex scene. The second set of experiments will study how people identify the shapes of objects moving in a scene. The third set of experiments will study how people judge the amount of depth between objects in the scene. These results will facilitate the three-stage development of the model. The first stage will use results from the first set of experiments to derive mathematically the number of independent objects that are moving in a computer-generated scene. The second stage will use results from the second set of experiments to describe mathematically the shape of small regions of each object. The third stage will use results from the third set of experiments to specify a quantitative value of the distance between objects in a scene. An understanding of how the visual system determines the shape and depth of multiple moving objects will have important implications for the design of computer-generated displays used to train pilots, for the development of visual aids for the blind, and for the development of robotics vision systems.