视觉皮层区功能柱型结构的认知计算意义  

The Significance of Cognitive Caculation Inspired by Parallel Model of Hypercolumns in Primary Visual Cortex

作　　者：危辉[1] 何新贵[2] 

机构地区：[1]北京航空航天大学计算机科学系,北京100083 [2]北京系统工程研究所,北京100101

出　　处：《计算机工程与应用》2000年第10期49-52,共4页Computer Engineering and Applications

基　　金：国家智能机研究开发中心并行处理机曙光1000的技术支持

摘　　要：在视皮层的初级皮层区有许多非常规整的柱形功能结构,它们形成的局域网络具有抽取视图象中最基本的微小特征的计算能力.相邻视神经细胞间的侧抑制机制和神经细胞的感受野为实现这样的并行计算能力提供了保证.并且视皮层的这种等级组构为许多心理现象提供了生理解释,对人工智能系统的知识获取和知识表示都具有非常重要的意义.该文设计的初级皮层视知觉并行计算模型就是以生理学依据为出发点的,取的是`解释智能'的思想.： In the primary visual cortex of our brain,there are a number of columnar functional organization regularly,each of which is a local neural network,the computation ability of them make the abstraction and segmentation of the critical features from the visual image possible,on these physiological fundament.And the receptive field of a neural cell and the mechanism of lateral inhibition among cells play a critical role in the signal processing in the brain.A lot of phenomena of psychology can be interpreted by this kind of hierarchical composition model of visual cortex.In this paper,a parallel hierarchical model was designed to acquit the orientation information from a visual image based on the hypercolumn structures in primary visual cortex．

关 键 词：神经视觉 知识获取 初级皮层区 人工智能 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]