摘要
溢油事故带来的海洋污染问题日益严重,SAR图像快速准确地自动识别为溢油事故的处理和决策支持提供了重要前提。为了获得更高的油膜识别准确率,提出了一种基于曲波变换(Curvelet)和局部线性嵌入(Local Linear Embedding,LLE)算法的SAR图像特征提取方法。首先,利用Curvelet对图像进行分解,选取包含了主要信息的低频分量作为新的图像矩阵;然后,利用LLE进行非线性降维,提取图像分类特征。为了验证提取特征的有效性,所提的Curvelet-LLE算法与PCA、LLE、等距特征映射(Isomap)、Curvelet变换和Fisher判别分析(Curvelet-KFD)、Wavelet-LLE等特征提取算法,利用K最近邻和支持向量机分类器分别进行了对比实验。实验结果表明,Curvelet-LLE算法能更有效地提取SAR图像油膜识别的分类鉴别特征,其准确率相对较高,具有较好的实用性。
The problem of marine pollution caused by oil spill accidents is becoming increasingly serious.The rapid and accurate automatic identification of synthetic aperture radar(SAR) images provides an important prerequisite for the handling and decision support of oil spill accidents.For the purpose of improving the precision rate of identification,a feature extraction method for SAR image is proposed based on the algorithm of Curvelet transform and local linear embedding(LLE).Firstly,the image is decomposed by Curvelet,and the low-frequency component containing the main information is selected as the new image matrix.Then the LLE is utilized for nonlinear dimensionality reduction for the data to extract the image classification features.In order to verify the validity of the extracted features,contrast experiments on the feature extraction algorithms such as PCA,LLE,Isomap,Curvelet-KFD and Wavelet-LLE are respectively carried on by the classifier of KNN and SVM.It turns out that the Curvelet-LLE can more effectively extract the features of oil spills identification in SAR image,with relatively higher accuracy and better practicability.
引文
[1] BREKKE C,SOLBERG A H S.Oil spill detection by satellite remote sensing[J].Remote Sensing of Environment,2005,95(1):1-13.
[2] SALBERG A B,RUDJORD O S. Oil spill detection in hybrid-polari metric SAR images[J].IEEE Transactions on Geoscience and Remote Sensing,2014,52(10):6521-6533.
[3] GUO Y,ZHANG H.Oil spill detection using synthetic aperture radar images and feature selection in shape space[J]. International Journal of Applied Earth Observation and Geoinformation,2014,30(8):146-157.
[4] CHAUDHURI D,SAMAL A,AGRAWAL A,et al.A statistical approach for automatic detection of ocean disturbance features from SAR images[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2012,5(4):1231-1242.
[5]周慧,陈澎.利用最小冗余最大相关和SVM的SAR图像海上溢油识别[J].电讯技术,2018,58(8):895-899.
[6] SADHAHBI S,BARHOUMI W,ZAGROUBA E. Breast cancer diagnosis in digitized mammograms using curvelet moments[J]. Computers in Biology and Medicine,2015,64(1):79-90.
[7]郭越,王晓峰.基于特征选取及广义傅立叶分形的SAR图像海洋溢油检测算法[J].海洋学报,2014,36(5):61-67.
[8]尹清婷.基于小波变换的SAR图像特征提取与识别研究[D].厦门:厦门大学,2013.
[9] CANDES E,DONOHO L.Continuous curvelet transform:I.resolution of the wavefront set[J]. Applied and Computational Harmonic,2005,19(2):162-197.
[10] BACHMANN C M,AINSWORTH T.Exploiting manifold geometry in hyperspectral imagery[J].IEEE Transactions on Geoscience and Remote Sensing,2005,43(3):441-454.
[11] CANDES E,DEMANET L,DONOHO D,et al.Fast discrete curvelet transforms[J]. Multiscale Modeling&Simulation,2006,5(3):861-899.
[12]贾西贝.基于曲波变换的人脸识别算法研究[D].哈尔滨:哈尔滨工程大学,2013.
[13] WU D,GUO H.Research on multi-feature fusion for discriminating oil spill and look-alike spots[C]//Proceedings of 2017 International Conference on Information Science and Control Engineering. Changsha:IEEE,2017:607-610.
[14]张琴,谷雨,徐英,等.混合智能优化算法的SAR图像特征选择[J].遥感学报,2016,20(1):73-79.
[15] ROWEI S T,SAUL L K.Nonlinear dimensionality reduction by locally linear embedding[J]. Science,2000,290(5500):2323-2326.
[16]陈澎.机载激光荧光海上溢油信息提取与反演研究[D].大连:大连海事大学,2012.
[17]张慧,肖蒙,崔宗勇.基于卷积神经网络的SAR目标多维度特征提取[J].机械制造与自动化,2017,46(1):111-115.