摘要
为提高极化合成孔径雷达(SAR)遥感图像对海面舰船目标的检测性能,提出了一种基于条件熵的多视极化SAR图像的舰船目标检测方法。首先将多视极化SAR数据进行Cloude特征分解,然后使用特征向量得到对应分量的相似性参数,利用特征值和相似性参数构造事件的概率和条件概率,从而构造功率条件熵。最后,该文提出了一种基于Parzen窗的杂波概率分布的估计方法,并基于该方法得到的概率分布提出了一种检测阈值的搜索方法。使用该方法,可以增加目标和杂波的对比度,并能准确搜索检测阈值。
A conditional entropy based ship detection method was developed to improve ship detection in multilook polarimetric synthetic aperture radar （PolSAR） images. First, Cloude decomposition is used for multilook PolSAR data, with the similarity parameter obtained from the eigenvectors. Then, the eigenvalues and the similarity parameter are used to construct a probability and a conditional probability to get the span conditional entropy. Finally, a Parzen window based method is used to estimate the probability distribution function （PDF） of the clutter followed by a search detection threshold method. This method improves the contrast between target and clutter and accurately sets the search detection threshold.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2012年第12期1693-1697,共5页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金资助项目(40871157
41171317)
关键词
极化合成孔径雷达
条件熵
PARZEN窗
polarimetric synthetic aperture radar
conditional entropy
Parzen window
