基于RBF神经网络的图像特征处理

Image Feature Extraction Based on RBFNN

径向基函数网络(RBFN)是当前人工神经网络技术研究的热点之一,并以其优良的性能广泛应用于各个领域。针对BP神经网络容易陷入局部最优的缺陷,文章提出用RBF神经网络进行图像特征提取,并给出了RBF神经网络构建过程及选Gauss函数为径向基函数的优势,最后设计并完成实验。模拟实验结果表明了训练好的RBF网络对边缘特征提取的有效性,也证实了选择良好的高斯函数宽度,可以得到较优的网络。

Radial Basis Function Network （RBFN） is a hot issue in the study of neural network, and is extensively applied into varied fields with its fine performance. Image Feature Extraction Based on RBFNN has been proposed in the paper in order to avoid the defects of BP neural network, at the same time, the details of the RBF neural network construction process and the advantage of chosing Gaus Funtion as RBF. The results showed its effectiveness, which proved that select a good Gaussian width, can be optimized network.