摘要
针对合成孔径雷达(SAR)目标识别问题,提出了一种有效的SAR图像预处理方法.首先通过自适应阈值分割、形态学滤波及几何聚类处理获得干净平滑的目标图像,再采用幂变换来增强图像质量,然后提取图像的主分量分析(PCA)、二维主分量分析(2DPCA)特征来进行识别.基于美国运动和静止目标获取与识别(MSTAR)计划录取的数据的实验结果表明,结合上述预处理,PCA,2DPCA的识别性能均可达到96.5%以上.
An efficient image pre-processing method is proposed for Synthetic Aperture Radar （SAR） image based automatic target recognition application. Firstly, the smoothed target image is segmented from the clutter background via adaptive threshold segmentation, morphological filter and geometric clustering processing. Secondly, power transformation is used to enhance the obtained target image. Finally, Principal Component Analysis （PCA） and 2-Dimensional Principal Component Analysis （2DPCA） features are extracted for classifying the target. Experimental results based on the Moving and Stationary Target Acquisition and Recognition （MSTAR） data show that the recognition performance of PCA and 2DPCA by using the proposed pre-processing method can reach more 96. 5%.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2007年第5期733-737,共5页
Journal of Xidian University
基金
国家自然科学基金资助(60302009)
教育部留学回国人员基金资助
关键词
合成孔径雷达
目标识别
主分量分析
二维主分量分析
synthetic aperture radar
targe recognition
principal component analysis
2-dimensional principal component analysis
