BINOCULAR MATCHING, DISPARITY CHANNELING, AND STEREOPSIS

拟议的项目涉及心理物理和计算两个方面。 人类立体视觉和双眼视觉的研究。到目前为止，大多数 计算研究主要集中在解决双目匹配问题上。 问题，而大多数心理物理学都集中在立体上 本地化或立体形式识别。这样做的总体目标是 这个项目是为了更好地了解立体视觉的过程 通过缓解这种不平衡。 拟议项目的具体目标是： 1)研究双眼配对的心理物理特性 人类视觉中的过程，并将这些特性与 立体定位。 2)探索人类视觉中视差通道的本质 系统。这包括：a)开发基于两者的计算模型 立体视觉的经典三通道理论和最新理论 基于更多的频道数量。 B)测试上述两种类型的心理物理模型 任务类型，涉及双目匹配的任务和 涉及双目匹配和涉及立体视觉的匹配 本地化。C)重新评估临床条件的含义 根据上述结果的立体盲性和立体异常 调查。 3)开发一种立体视觉的计算模型，该模型包含 前端在生物学上是合理的，这可以使量化 关于两种类型的心理物理实验的预测， 涉及双目匹配(例如相关检测)和 涉及立体定位(例如，立体视觉)。 心理物理实验将涉及两种任务中的一种。 在第一类任务中，观察者将被要求做出判断 关于动态随机元的点间相关性 立体图，将在微型计算机上生成和显示。 不需要对视差或深度进行判断。在 第二类任务，观察者将被要求做出判断 在几乎相同的刺激下涉及立体定位 条件。不同刺激变量对脑电信号的不同影响 这两种类型的判决将被分析以区分 由双眼机制控制的立体视觉特性 与那些由立体机制支配的匹配 本地化。

The proposed project involves both psychophysical and computational investigations of human stereopsis and binocular vision. To date, most computional investigations have focused on solving the binocular matching problems while most of the psychophysics have focused on stereoscopic localization or stereoscopic form recognition. The broad goal of this project is to gain a greater understanding of the process of stereopsis by alleviating this imbalance. The specific aims of the proposed project are: 1) To investigate the psychophysical properties of the binocular matching process in human vision and to compare and contrast these properties with those of stereoscopic localization. 2) To explore the nature of disparity channeling in the human visual system. This includes: a) Developing computational models based on both the classical 3-channel theories of stereopsis and more recent theories based on larger number of channels. b)Testing both of the above types of models psychophysical using two types of tasks, those that involve binocular matching and those that involve binocular matching and those that involve stereoscopic localization. c) Reassessing the meaning of the clinical conditions of stereoblindness and stereoanomaly in light of the results of the above investigations. 3) To develop a computational model of stereopsis that incorporates a biologically plausible front end and that can make quantitative predictions about two types of psychophysical experiments, those involving binocular matching (e.g. correlation detection) and those involving stereoscopic localization (e.g. stereoacuity). The psychophysical experiments will involve one of two kinds of tasks. In the first type of task, observers will be required to make judgments concerning the interoccular correlation of dynamic random element stereograms, which will be generated and displayed on a microcomputer. No judgments concerning disparity or depth will be required. In the second type of task, observers will be required to make judgments involving stereoscopic localization under nearly identical stimulus conditions. The differential effects of various stimulus variables on these two types of judgments will be analyzed to distinguish the properties of stereopsis that are governed by the mechanisms of binocular matching from those that are governed by the mechanisms of stereoscopic localization.