摘要
通过分析传统鉴别分析的"小样本"和"次优性"问题,提出一种改进的二维线性鉴别分析(I2DLDA)算法并用于SAR图像目标特征提取。首先对线性鉴别分析中散度矩阵的构造进行加入权值的改进以缓解次优性问题,然后使用二维线性鉴别分析准则在图像矩阵上进行特征提取。对美国MSTAR计划公开的SAR图像数据的仿真实验结果表明,基于I2DLDA的SAR目标识别方法不仅有效增强了提取特征的可鉴别性,同时也减小了所需的特征维数,降低了运算量,识别性能有了很大的提高,证明了方法的有效性。
By analyzing the ＂small sample size＂ （SSS） problem and the ＂inferior＂ problem of traditional linear discriminant analysis （LDA）, a novel image feature extraction technique is proposed which is called improved two - dimensional linear discriminant analysis （I2DLDA）. Firstly, the scatter matrices in the discriminant analysis are modified by weighting to relieve the ＇ inferior＇ problem. Then, feature matrix is extracted in the image matrix by two-dimensional linear discriminant analysis criterion. Experimental results with MSTAR dataset show that the discrimination of feature is strengthened and also the feature dimensionality and computation complexity are reduced according to the recognition method based on I2DLDA. The better recognition performance demonstrates the effectiveness of the method.
出处
《电子信息对抗技术》
2013年第2期19-23,共5页
Electronic Information Warfare Technology
