The perception of 3D shape from texture

从纹理感知 3D 形状

• 批准号: 0546107
• 负责人: James Todd
• 金额: --
• 依托单位: Ohio State University Research Foundation -DO NOT USE
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-04-15 至 2011-03-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0546107&HistoricalAwards=false
• 关键词: perception; 3D; shape; texture

Our eyes present to us a world in three dimensions, and yet the information coming in through the eyes is not three dimensional. The retina that sits at the back of each eyeball is effectively a two-dimensional sheet of receptor cells. The light that is reflected or emitted from three-dimensional objects in the world is actually transduced as two-dimensional patterns on the retina. So how do we perceive the third dimension of depth on the basis of these two-dimensional patterns? This is a classic question in the area of visual perception, and many decades of research have resulted in a list of depth cues that our visual systems use to "reconstruct" the third dimension. What we are still trying to understand is, precisely how does the visual system utilize each cue to perform this feat of perception?With support of the National Science Foundation, Dr. Todd will closely examine the nature of one visual depth cue called "optical texture". To illustrate, imagine looking closely at a dimpled golf ball. The visual appearance of the dimples is warped by the curvature of the ball, and our visual systems can use this warping as a source of information about its roundness in all three dimensions. Dr. Todd has developed a computational model of how the visual system calculates the shape of an object based on its textural patterning, and the present research project will investigate whether his model is an accurate characterization of how the human visual system uses optical texture as a depth cue. Experiments will be conducted to precisely test how human participants use optical texture, and the results from these experiments will be compared against the predictions of the model. Besides advancing our knowledge of human visual perception, these results may inform the design of more robust and effective algorithms in machine vision and computer graphics, and more functional prosthetic devices for the blind.

我们的眼睛在三个维度上向我们展示了一个世界，但是通过眼睛传来的信息并不是三维。 位于每个眼球背面的视网膜有效地是二维受体细胞。 实际上，从世界上三维物体中反映或发出的光实际上被视为视网膜上的二维模式。 那么，我们如何根据这些二维模式来感知深度的三个维度？ 这是视觉感知领域的一个经典问题，数十年的研究导致了我们的视觉系统用来“重建”第三维的深度线索。 我们仍在尝试理解的是，恰恰是视觉系统如何利用每个提示来执行这一感知的壮举？在国家科学基金会的支持下，托德博士将仔细研究一种视觉深度提示的性质，称为“光学纹理”。 为了说明，想象一下仔细观察一个斑驳的高尔夫球。 凹痕的视觉外观是通过球的曲率翘曲的，我们的视觉系统可以将此翘曲用作有关其在所有三个维度中其圆度的信息来源。 Todd博士开发了一个计算模型，该模型是，视觉系统如何根据对象的纹理图案计算对象的形状，而本研究项目将研究他的模型是否是对人类视觉系统如何使用光学纹理作为深度提示的准确表征。 将进行实验，以精确测试人类参与者如何使用光学纹理，并将这些实验的结果与模型的预测进行比较。 除了促进我们对人类视觉感知的了解外，这些结果还可以为机器视觉和计算机图形方面的更健壮和有效的算法以及盲人的功能性假肢设备提供信息。