基于改进的粒子群径向基神经网络的目标识别  

Target recognition based on radial basis function neural network optimized by modified particle swarm optimization

作　　者：袁艳[1] 叶俊浩 苏丽娟[1] YUAN Yan;YE Junhao;SU Lijuan(School of lnstrumentation Science and Opto-Electronics Engineering,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学仪器科学与光电工程学院,北京100191

出　　处：《计算机应用》2018年第A01期6-8,23,共4页journal of Computer Applications

基　　金：国家自然科学基金重点项目(61635002)

摘　　要：为了提高径向基(RBF)神经网络对航拍影像目标的识别率,提出了一种权重改进的粒子群优化(PSO)算法优化径向基神经网络,进行目标识别。首先,运用权重改进的PSO算法求解RBF神经网络隐含层中心,获取优化的径向基神经网络的权值和阈值;合理地选择待识别目标的样本图像;最后,采用训练过的径向基神经网络对航拍疑似目标图像进行识别。分别采用该算法、经正交最小二乘(OLS)算法和基本PSO算法优化的RBF神经网络对航拍影像进行疑似目标提取和识别,实验结果表明,所提算法对隐含层节点较少的RBF神经网络,识别正确率达到98%,识别效果最好。In order to improve Radial Basis Function （RBF） neural network recognition rate of airborne images, a RBF nem＇al network optimized by Particle Swarm Optimization （PSO） algorithm with improved weight was proposed for target recognition. Firstly, the PSO algorithm was used to generate the hidden layer centers of RBF neural network, the weights and thresholds of the optimized RBF neural network were obtained. The sample images with objects to be recognized were selected reasonably. Finally, the suspected target images were recognized by the trained RBF neural network. The experimental results indicate that comparing with the RBF neural network optimized by Orthogonal Least Squares （OLS） algorithm and basic PSO algorithm, the correct recognition rate of the proposed algorithm is higher to 98% while the RBF neural network has fewer hidden layer nodes.

关 键 词：粒子群优化算法 径向基 神经网络 识别 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]