人类视觉系统超视锐度现象的神经网络数学模型研究

超视锐度是人们熟知的奇特现象,虽然许多视觉研究领域中的专家们对此做了大量的研究,但视觉系统究竟是如何提取超视锐度信息的却仍然是个未解决的问题.本文用简单细胞广义Gabor函数为基本功能单元,以Marr提出的计算理论为框架,建立了一个感受野重叠的,具有比较运算性质的、可解释超视锐度现象的神经网络数学模型.它在空间比较,真实运动和似动三方面较好地描述了已有的一些有关超视锐度现象的心理物理实验结果.

A MATHEMATICAL MODEL OF NEURAL NETWORK EXPLAINING HYPERACUITY IN THE HUMAN VISUAL SYSTEM

The phenomenon of hyperacuity has been a well-known and strange one in the human visual system. Though a great deal of investigations on it have been done, what and how the neural mechanism in the visual system processes the information of hyperacuity still remain open. In this paper, we use Extended Cabor Function (EG) of simple cells as the basic functional units to have successfully set up a mathematical model of neural network, with the properties of overlap-formed receptive fields and of comparative operation, explaining hyperacuity. Our computer simulating results indicate that this model can well explain the psychophysical experiments of hyperacuity in three conditions: spatial comparison, real motion and apparent motion.