Surface smoothing v.s. boundary detection in the perception of 3D scenes

表面平滑与表面平滑

• 批准号: 203634-2007
• 负责人: Wilcox, Laurie
• 金额: $ 1.66万
• 依托单位: York University
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2008
• 资助国家: 加拿大
• 起止时间: 2008-01-01 至 2009-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=392887
• 关键词: Surface; smoothing; v.s.; boundary; detection

The natural environment contains smooth surfaces defined by edges that define where one object, or surface ends, and the other begins. Many experiments have shown that human visual system is adept interpolating such surfaces using a single cue to depth, stereopsis. However, stereopsis is also extremely useful in providing very precise estimates of relative depth between two regions. How does the visual system resolve the conflict between smoothing depth estimates across surfaces, and detecting the depth discontinuities between them? The hypothesis that is central to this proposal is that the brain uses unmatched elements along an object boundary to signal that an edge is present, and so define a discontinuity that cannot, or should not, be smoothed.

自然环境包含由边定义的平滑曲面，这些边定义了一个对象或曲面的结束位置和另一个对象或曲面的开始位置。大量的实验表明，人类视觉系统能够很好地利用单一线索对曲面进行插值，从而获得深度、立体感。然而，立体视觉在提供两个区域之间的相对深度的非常精确的估计方面也非常有用。视觉系统如何解决平滑表面上的深度估计和检测它们之间的深度不连续性之间的冲突？这一提议的核心假设是，大脑使用沿物体边界沿着不匹配的元素来发出边缘存在的信号，从而定义不能或不应该被平滑的不连续性。