Kitazaki M, Shimojo S
Department of Life Sciences/Psychology, University of Tokyo, Japan.
Perception. 1996;25(7):797-814. doi: 10.1068/p250797.
The generic-view principle (GVP) states that given a 2-D image the visual system interprets it as a generic view of a 3-D scene when possible. The GVP was applied to 3-D-motion perception to show how the visual system decomposes retinal image motion into three components of 3-D motion: stretch/shrinkage, rotation, and translation. First, the optical process of retinal image motion was analyzed, and predictions were made based on the GVP in the inverse-optical process. Then experiments were conducted in which the subject judged perception of stretch/shrinkage, rotation in depth, and translation in depth for a moving bar stimulus. Retinal-image parameters-2-D stretch/shrinkage, 2-D rotation, and 2-D translation-were manipulated categorically and exhaustively. The results were highly consistent with the predictions. The GVP seems to offer a broad and general framework for understanding the ambiguity-solving process in motion perception. Its relationship to other constraints such as that of rigidity is discussed.
一般视图原则(GVP)指出,对于给定的二维图像,视觉系统会尽可能将其解释为三维场景的一般视图。GVP被应用于三维运动感知,以展示视觉系统如何将视网膜图像运动分解为三维运动的三个分量:拉伸/收缩、旋转和平移。首先,分析了视网膜图像运动的光学过程,并在逆光学过程中基于GVP进行了预测。然后进行了实验,让受试者判断对移动条形刺激的拉伸/收缩、深度旋转和深度平移的感知。视网膜图像参数——二维拉伸/收缩、二维旋转和二维平移——被分类且详尽地进行了操控。结果与预测高度一致。GVP似乎为理解运动感知中的歧义解决过程提供了一个广泛而通用的框架。文中还讨论了它与其他约束条件(如刚性约束)的关系。