一种改进的Zernike矩的雷达图像目标识别方法  

An Improved Radar Images Target Recognition Based on Zernike Invariant Moment

作　　者：石祥滨[1] 刘东琦[1] 于泳浩[1] 

机构地区：[1]辽宁大学信息学院,辽宁沈阳110036

出　　处：《辽宁大学学报（自然科学版）》2012年第2期160-164,共5页Journal of Liaoning University：Natural Sciences Edition

基　　金：辽宁省科技厅资助课题

摘　　要：针对Zernike不变矩目标识别算法计算复杂,计算复杂度高的不足和雷达图像乘性噪声严重和识别率不高的问题,提出了一种快速计算Zernike矩目标识别算法.该算法通过利用Zernike矩对称性和反对称性对雷达图像进行快速计算求得不变矩,然后通过加权求得新的特征向量,最后通过支持向量机进行目标识别.实验结果表明,提出的算法不仅解决了计算过于复杂的问题,还降低了噪声且提高了目标识别率.In order to resolve computational complexity for the Zernike Moment Invariant target recognition algorithm,and the recognition rate of radar images is not high,a quick calculation of the Zernike moments target recognition algorithm is put forward.The algorithm through the use of the symmetry of the Zernike moments of radar images quickly calculable in the same moment,and then weighted to obtain a new feature vector,the final target recognition by support vector machine.Theexperimental results show that the proposed algorithm not only solved the problem of calculation is too complex,also reduces the noise and improve the target recognition rate.

关 键 词：目标识别 Zernike不变矩 特征提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]