【摘要】 提出一种基于径向基(RBF)函数神经网络的曲线重构学习方法,即由描述物体轮廓特征的样本点作为RBF神经网络的学习样本,利用RBF神经网络强大的函数逼近能力对样本点进行学习和训练,从而仿真出包含这些样本点的原始曲线,同时对于曲线一些样本点缺少的情况下,仍然能构通过调整参数训练得到这些样本点的原始拟和曲线。实验表明,基于径向基(RBF)函数的神经网络具有很强的物体边界描述能力和缺损修复能力。更多还原

【Abstract】 A new method based on radial basis function(RBF) neural network for curve reconstruction is introduced.The sample which can describe curve’s feature used as input of the RBF neural network,then take advantage of RBF’s powerful ability of function approach to learn and train these sample,thus constructs the curve which contains these sample.Meanwhile,when sample is not enough to describe a curve,this method can also construct a fitting curve and satisfy the precision.Experimental results testify that RBF neural network has a powerful ability on construting and restoring a curve.更多还原