摘要
将一种动态递归神经网络完成的最近邻分类器(NNC)应用于彩色图像恢复。采用多层感知器(WLP)与径向基函数(RBF)网络相结合的网络结构,把原型模式的显式表示作为网络参数,可自由扩大或删除原型模式,具有自适应特性。用这种模型实现的动态NNC去除了传统NNC中的比较运算,一定程度上降低了运算复杂度。试验结果表明该方法对于含有不同程度噪声的彩色图像恢复效果良好。
Because the dynamic recurrent Neural Network that performs nearest neighbor classification shows good performance, it was adopted for color image restoration in this paper. It combines a MLP and a Radial Basis Function (RBF) Neural Network, and allows for explicit representation of prototype patterns as network parameters, as well as adding or deleting prototype freely, so the structure of this system is characterized by self-adaptation. The dynamic classification implemented by the network eliminates all comparisons, which are the vital steps of the conventional Nearest Neighbor Classification (NNC), and to some degree it is also simple for computation. The results show this model restores the typically degenerated image with different noise ratios excellently.
出处
《计算机应用》
CSCD
北大核心
2007年第5期1160-1163,共4页
journal of Computer Applications
关键词
径向基函数
神经网络
图像恢复
最近邻分类器
Radial Basis Function(RBF)
neural network
image restoration
the Nearest Neighbor Classification(NNC)