一种基于神经网络的彩色图像分割方法  

Color Image Segmentation based on the Neural Network

作　　者：张桂敏[1] 汪熙[1] 要俊杰[1] 程高峰[1] 

机构地区：[1]装甲兵工程学院,北京100072

出　　处：《火力与指挥控制》2008年第S2期22-23,38,共3页Fire Control & Command Control

摘　　要：提出了适于彩色图像分割的加强型径向基函数网络方法,采用在线自适应聚类学习算法确定隐层节点的数目、激活函数中心值;通过Hebb学习算法迅速将隐层节点中心值分为目标颜色聚类中心和背景颜色聚类中心两类;输出层用竞争规则将目标与背景分开。通过对多幅彩色图像进行分割处理验证了该方法的有效性。A new neural network method,which is called an enhancing learning on the radial basis function neural network,is presented to perform color image segmentation.A on-line self-adaptive clustering algorithm is employed to set the hidden layer and select the radial basis function center as well.The Hebb learning algorithm is introduced to train the hidden layer in order to divide its hidden neurons center vectors into two meaningful groups: one group members are the object color-clustering centers and the others are the background centers.The competitive algorithm is introduced to train the output layer,which puts out different values with different input values,hence divides the object pixels from the image.Experiments are executed on many color images and the results exhibit the proposed approach is efficient.

关 键 词：彩色图像分割 RBF网络 ERBF网络 Hebb算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]