Depth perception in normal and abnormal vision

正常和异常视力的深度知觉

• 批准号: 9912390
• 负责人: Jian Ding
• 金额: $ 36.88万
• 依托单位: UNIVERSITY OF CALIFORNIA BERKELEY
• 依托单位国家: 美国
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-02-01 至 2023-12-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9912390
• 关键词: 3-Dimensional; Affect; Amblyopia; Anisometropia; Binocular Vision; Brain; Complex; Computer Models; Contrast Sensitivity; Depth Perception; Discrimination; Eye; Frequencies; General Population; Goals; Human; Image; Individual; Lesion; Life; Measures; Methods; Modeling; Nature; Neurons; Noise; Perception; Phase; Positioning Attribute; Property; Resources; Running; Signal Transduction; Specific qualifier value; Stimulus; Strabismus; Structure; System; Testing; Three-Dimensional Image; Time; Vision; Vision Disparity; Visual; Visual Acuity; Visual system structure; base; blind; developmental disease; experimental study; luminance; models and simulation; monocular; novel; receptive field; response; spatial vision; spatiotemporal; three-dimensional modeling

Project Summary The human brain devotes enormous resources toward providing a cyclopean view of the world, by combining the separate inputs from the two eyes. The resulting binocular vision has substantial benefits in the form of stereopsis. In addition to the well-documented monocular deficits, humans with amblyopia (a developmental disorder affecting 2-3% of the general population) also suffer abnormal depth perception, and many more individuals are stereo-blind. However, the binocular deficits in amblyopia have seldom been assessed over the whole range of binocular disparity. In this project, we propose to use a novel rating-scale method to evaluate depth perception over the whole range of binocular disparities, and to use dynamic bandpass noise stereograms with depth corrugation to reveal spatial properties of depth profiles, and to develop and test a new model to predict depth perception pixel-by-pixel over a very broad range of stimulus conditions. The long-term objectives of this project are to understand how the brain combines the two eyes' inputs to form 3D images in humans with normal and abnormal binocular vision. We propose to develop a novel 3D model by combining a disparity sensitivity model with our previous 2D model. Aim 1: Mechanisms of depth perception. We propose to develop and test a new depth perception model with a filter/binocular-energy/filter (F-BE-F) structure, to predict depth perception pixel by pixel over a very broad range of conditions. The two eyes images pass through first-stage spatiotemporal filters to calculate the normalized binocular-energy (BE), which goes through a maximum (MAX) operator for solving the correspondence problem, then goes through a disparity window to compute local depth quantities, and finally through a second stage of spatiotemporal filtering to give the final perceived depth profile. Different mechanisms of depth perception in human vision can be isolated by different normalizations of the BE and different spatiotemporal properties of depth perception. Aim 2: Spatiotemporal properties of depth perception. Our preliminary experiments suggest that depth perception is dependent on stimulus size and duration, which can be explained by a second stage of spatiotemporal filtering. We will further study these spatiotemporal properties of depth perception (a) at different stimulus spatial frequencies; (b) at different stimulus orientations; (c) at different pedestal disparities in the reference background. (d) We will perform both experiments and model simulations to extract form information from stereograms with varying spatiotemporal depth profiles. Aim 3: Abnormal depth perception in amblyopia. We will measure and model depth perception in humans with abnormal binocular vision due to amblyopia. We will measure depth perception over a large range of conditions, simulating abnormalities both empirically (by stimulus manipulation), and theoretically, by making `lesions' in our model, in order to determine the nature and cause of reduced stereopsis in amblyopia.

项目摘要 人类的大脑投入了巨大的资源，通过结合 两只眼睛的不同输入。由此产生的双目视觉具有以下形式的实质性益处： 立体视觉除了记录良好的单眼缺陷外，患有弱视的人（一种发育缺陷） 影响2-3%的普通人群的疾病）也患有异常的深度知觉，等等 个体是立体盲。然而，弱视的双眼缺陷很少被评估， 双眼视差的全范围。在这个项目中，我们建议使用一种新的评级尺度方法来评估 深度知觉在整个范围内的双眼差异，并使用动态带通噪声 立体图与深度剖面图，以揭示深度剖面的空间特性，并开发和测试一种新的 模型来预测深度知觉像素像素在一个非常广泛的刺激条件。长期 该项目的目标是了解大脑如何将两只眼睛的输入结合起来形成3D图像， 具有正常和异常双眼视觉的人类。我们建议开发一种新的3D模型， 视差敏感度模型与我们以前的2D模型。目的1：深度知觉的机制。我们 建议开发和测试一个新的深度感知模型与过滤器/双目能量/过滤器（F-BE-F） 结构，在非常广泛的条件下逐像素地预测深度感知。两只眼睛的图像 通过第一级时空滤波器以计算归一化的双目能量（BE），其 通过用于解决对应问题的最大值（MAX）算子，然后通过视差 窗口计算局部深度量，最后通过第二阶段的时空滤波， 最终的感知深度轮廓。人类视觉中深度感知的不同机制可以通过以下方式来分离： BE的不同归一化和深度感知的不同时空特性。目标二： 深度知觉的时空特性。我们的初步实验表明， 是依赖于刺激的大小和持续时间，这可以解释为第二阶段的时空 过滤我们将进一步研究这些时空特性的深度知觉（a）在不同的刺激 空间频率;（B）在不同的刺激方向;（c）在不同的基座差异的参考 背景(d)我们将进行实验和模型模拟，以从 具有变化的时空深度轮廓的立体图。目的3：弱视深度觉异常。 我们将测量和模拟人类的深度知觉异常的双眼视觉由于弱视。我们 将在大范围的条件下测量深度知觉，根据经验（通过 刺激操纵），理论上，通过在我们的模型中制造“病变”，以确定性质和 弱视的立体视觉降低的原因。