摘要
文章提出了一种基于视觉模型的图像边缘检测算法．作者采用具有反馈和前项控制连接的侧抑制模型进行图像信息进行处理，用零阶和二阶厄米特（Ｈｅｒｍｉｔｅ）函数的组合产生系统的控制模板，组合系数通过梯度下降学习算法确定．这种以生物视觉感知机理为基础的神经计算方法可有效地均衡滤除噪声和增强边缘，同时将系统高维参数 的学习问题转化为对少数几个参数的确定，降低了问题的维数和计算的复杂性．
An algorithm of edge detection based on visual model is Proposed. In which image information is processed by using a lateral inhibitory modal with feedback and forward coned connections. The forward templates are generated by a combination of zero- and Second- Order Hermite functions with scale parameters. Corresponding combination coefficients are trained by neural computing. Such method based on biological visual mechanisms can efficiently perform balancing of noise reduction and edge enhancement. Furthermore,the dimension and complexity of Problems are reduced.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2000年第1期101-103,共3页
Acta Electronica Sinica
基金
国家自然科学基金!(NO.69735010)
关键词
边缘检测
视觉模型
图像处理
算法
the detection
Hermite function
visual model
neural computing
