摘要
提出了一个早期视觉的可能的计算模型,把几何特征区分为局部特征与全局特征,认为早期视觉中感知局部几何性质和全局拓朴性质先于运动感知,先于形状感知.模型本质上虽然仍然是计算的,但已不是传统的vonNeumann的顺序计算,而是非vonNeumann的平行计算,它为计算理论与拓扑性质检测问题之间的矛盾提供了一个可能的解决方法.
A computational model for early vision is proposed to support the hypothesis of early perception of topological properties. This model suggests a sequence of early visual perception: preprocessing local geometric primitives and global topological primitives extraction,motion perception, and shape (global geometric properties) perception. The computational nature of this model provides a possible solution to the contradiction between the topological perception and the computational approaches to vision.
出处
《复旦学报(自然科学版)》
CAS
CSCD
北大核心
1997年第6期601-608,共8页
Journal of Fudan University:Natural Science
基金
攀登计划项目
关键词
整体性质
拓扑性质
视知觉
计算机视觉
global properties
geometric properties
topological properties
visual perception
early vision