合成孔径雷达（Synthetic Aperture Radar，SAR）图像变化检测的目的是分析和处理相同区域内不同时间的多幅SAR图像，从而获取区域内的变化信息，它被广泛地用于灾害评估，环境监测，森林监测以及战场形势分析。因此，本论文以SAR图像为研究对象，以稀疏表示为理论基础，针对传统变化检测技术在凭经验融合高维度特征信息所导致特征冗余和过度拟合问题，展开基于多特征提取与最优选择的SAR图像变化检测算法研究。本文的主要研究内容概括如下：

（1）提出了一种基于稀疏表示算法的SAR图像特征筛选及变化检测方法，该算法根据基于稀疏系数的特征筛选得到的重构信息来表征原样本，即考虑到每个特征携带信息量的不同，将稀疏表示算法与皮尔逊相关系数法相结合，挖掘信息量比较丰富的特征向量，利用稀疏表示进行局部重构，表征之前的特征样本，进而实现特征筛选，然后输入支持向量机（Support Vector Machine，SVM）分类器中，最终得到变化检测结果。实验结果表明，与基于主成分分析、稀疏主成分分析的降维方法相比，稀疏表示在SAR图像变化检测中得到了有效的应用，其变化检测精度有了很大的改善。

（2）在SAR图像变化检测中，SAR图像的空间特征信息、空间邻域信息和统计分布特性起着至关重要的作用，针对多特征决策融合的SAR图像变化检测问题，本文提出了基于联合稀疏表示的特征筛选及变化检测方法，该方法采用稀疏表示算法分别求解各类特征的稀疏系数矢量，以及确定两个系数向量间的相关关系；其次，通过构建互相关矩阵，求解非线性相关信息熵；最后，根据最大熵原则，获得最佳的特征组合，并进行联合表征。实验结果表明，与稀疏表示算法相比，联合稀疏表示算法能够对SAR图像信息进行综合高效的融合，其抗噪性、鲁棒性以及边界保持性等方面都更为优越，可以获得更好的变化检测精度。

Synthetic Aperture Radar (SAR) change detection refers to the analysis of SAR image changes at different time points in the same region, so as to judge the scene change information and its process in the region. It is widely used in the fields of national economy and national defense construction such as disaster assessment, environmental monitoring, forest monitoring and battlefield situation analysis. In view of the phenomenon of feature redundancy and overfitting caused by feature selection only based on experience when traditional change detection technology fuses high-dimensional feature information of SAR images, this paper takes spaceborne SAR images acquired at different times in the same area as the basis and sparse representation theory as the basis to carry out research on relevant technical issues of SAR image change detection algorithm. Specific research contents are as follows:

(1) A SAR image feature screening and change detection method based on sparse representation algorithm is proposed. This algorithm represents the original sample according to the reconstructed information obtained from feature screening based on sparse coefficient. That is, considering the different information carried by each feature, the sparse representation algorithm is combined with Pearson correlation coefficient method to excavate feature vectors with relatively rich information, obtain sparse representation local reconstruction, and characterize the previous feature samples. Input into SVM, and finally get the change detection result. The simulation results show that, compared with the dimensionality reduction methods based on principal component analysis and sparse principal component analysis, sparse representation is effectively applied in SAR image change detection, and its change detection accuracy has been greatly improved.

(2) In the SAR image change detection, the spatial feature information, spatial neighborhood information and statistical distribution characteristics of SAR images play a crucial role. Aiming at the SAR image change detection problem of multi-feature decision fusion, this paper proposes a feature screening and change detection method based on joint sparse representation. The sparse representation algorithm is used to solve the sparse coefficient vectors of various features and determine the correlation between the two coefficient vectors. Then, by constructing the cross-correlation matrix, the nonlinear correlation information entropy is solved. Finally, according to the principle of maximum entropy, the best combination of features is obtained and the joint characterization is carried out. The experimental results show that, compared with the sparse representation algorithm, the joint sparse representation algorithm can integrate SAR image information comprehensively and efficiently, and its anti-noise, robustness and boundary retention are superior, which can obtain higher change detection accuracy.

[1] Sun H, Shimada M, Feng X. Recent Advances in Synthetic Aperture Radar Remote Sensing-Systems, Data Processing, and Applications[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(11): 2013-2016.

[2] Li M, Li M, Zhang P, et al. SAR Image Change Detection Using PCANet Guided by Saliency Detection[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 16(3): 402-406.

[3] Pirrone D, Bovolo F, Bruzzone L. A Novel Framework Based on Polarimetric Change Vectors for Unsupervised Multiclass Change Detection in Dual-Pol Intensity SAR Images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(7): 4780-4795.

[4] Li L, Wang C, Zhang H, et al. Urban Building Change Detection in SAR Images Using Combined Differential Image and Residual U-Net Network[J]. Remote Sensing, 2019, 11(9): 1091.

[5] Fang W, Xi C. Land-Cover Change Detection for SAR Images Based on Biobjective Fuzzy Local Information Clustering Method With Decomposition[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 1-5.

[6] Saha S, Bovolo F, Bruzzone L. Building Change Detection in VHR SAR Images via Unsupervised Deep Transcoding[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 59(3): 1917-1929.

[7] 梅妍玭,张得才,傅荣.基于形态学与多尺度空间聚类的SAR图像变化检测方法研究[J].光电子·激光,2021,32(11):1140-1146.

[8] 袁晓谦,陈超,田姗.基于邻域信息的SAR图像变化检测[J].激光杂志,2021,42(01):118-123.

[9] Chen L C, Papandreou G, Kokkinos I, et al. DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(4): 834-848.

[10] Taskin G, Kaya H, Bruzzone L. Feature Selection Based on High Dimensional Model Representation for Hyperspectral Images[J]. IEEE Trans Image Process, 2017, 26(6): 2918-2928.

[11] 何兴高,李蝉娟,王瑞锦,等.基于信息熵的高维稀疏大数据降维算法研究[J].电子科技大学学报,2018,47(02):235-241.

[12] Xu B, Ding X, Hou R, et al. A Feature Extraction Method Based on Stacked Denoising Autoencoder for Massive High Dimensional Data[C]// 2018 14th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD). IEEE, 2018: 206-210.

[13] Wang R, Zhang J, Chen J, et al. Imbalanced Learning-Based Automatic SAR Images Change Detection by Morphologically Supervised PCA-Net[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 16(4): 554-558.

[14] Chen H, Wu C, Du B, et al. Change Detection in Multisource VHR Images via Deep Siamese Convolutional Multiple-Layers Recurrent Neural Network[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(4): 2848-2864.

[15] Pirrone D, Bovolo F, Bruzzone L. A Novel Framework Based on Polarimetric Change Vectors for Unsupervised Multiclass Change Detection in Dual-Pol Intensity SAR Images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(7): 4780-4795.

[16] Xue D, Lei T, Jia X, et al. Unsupervised Change Detection Using Multiscale and Multiresolution Gaussian-Mixture-Model Guided by Saliency Enhancement[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14: 1796-1809.

[17] Luppino, Luigi Tommaso, Bianchi, et al. Unsupervised Image Regression for Heterogeneous Change Detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(12): 9960-9975.

[18] Li A, Haraguchi M, Okubo Y, et al. Finding What Changes for Two Graphs Constructed from Different Time Intervals[C]// 2012 IIAI International Conference on Advanced Applied Informatics (IIAIAAI). IEEE, 2012: 48-53.

[19] 储艳丽.基于支持向量机的遥感图像变化检测[J].信息技术,2016,40(10):148-151.

[20] Gong M, Zhan T, Zhang P, et al. Superpixel-Based Difference Representation Learning for Change Detection in Multispectral Remote Sensing Images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(6): 2658-2673.

[21] 黄晨霞,殷君君,杨健.基于L1范数主成分分析的极化SAR图像变化检测[J].系统工程与电子技术,2019,41(10):2214-2220.

[22] 刘梦岚.基于特征融合的SAR图像变化检测[D].合肥:合肥工业大学,2019.

[23] Guo X, Yin J, Yang J. Comparative Analysis of Change Detection Methods in Polarimetric SAR Images Based on Probability Statistical Models[C]//2021 CIE International Conference on Radar (Radar). IEEE, 2021: 604-609.

[24] 朱磊,李敬曼,潘杨,等.自适应调节滤波强度的SAR图像非局部平均抑斑算法[J].电子与信息学报,2021,43(05):1258-1266.

[25] Alkandari A, Aljaber S J. Principle Component Analysis Algorithm (PCA) for Image Recognition[C]// 2015 Second International Conference on Computing Technology and Information Management (ICCTIM). IEEE, 2015: 76-80.

[26] Seghouane A K, Shokouhi N, Koch I. Sparse Principal Component Analysis With Preserved Sparsity Pattern[J]. IEEE Transactions on Image Processing, 2019, 28(7): 1-1.

[27] Jolliffe, Ian T. Rotation of Principal Components: Choice of Normalization Constraints[J]. Journal of Applied Stats, 2012, 22(1): 29-35.

[28] Matematica J C D D, Ian T. Jolliffe Department of Mathematical Sciences. Loading and Correlations in The Interpretation of Principle Compenents[J]. Journal of Applied Stats, 1995, 22(2): 203-214.

[29] Seghouane A K, Shokouhi N, Koch I. Sparse Principal Component Analysis with Preserved Sparsity Pattern[J]. IEEE Transactions on Image Processing, 2019, 28(7): 3274-3285.

[30] 李亚娟.结合全局和局部稀疏表示的SAR图像目标识别方法[J].电子测量与仪器学报,2020,34(02):165-171.

[31] 张仁霖.SRC人脸识别算法分析及改进[J].唐山师范学院学报,2020,42(06):74-77.

[32] Shekhar S, Patel V M, Nasrabadi N M, et al. Joint Sparse Representation for Robust Multimodal Biometrics Recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(1): 113-126.

[33] 李荷镜.基于条件随机场的SAR图像变化检测算法研究[D].西安:西安电子科技大学,2014.

[34] 田元荣,许悦雷,田松,等.基于稀疏表示模型的SAR图像目标检测算法[J].中国科技论文,2013,8(10):1025-1028+1034.

[35] 黄博,周劼,江舸,等.高分辨SAR目标复杂结构特征增强成像算法[J].红外与毫米波学报,2022,41(04):762-769.

[36] Kwak Y, Song W J, Kim S E. Speckle-Noise-Invariant Convolutional Neural Network for SAR Target Recognition[J]. IEEE Geoscience and Remote Sensing Letters, 2019, 16(4): 549-553.

[37] Singh. Digital Change Detection Techniques Using Remotely Sensed Data[J]. International Journal of Remote Sensing, 1989, 10(6): 989-1003.

[38] Yong L, Zhu D. The Geometric-Distortion Correction Algorithm for Circular-Scanning SAR Imaging[J]. IEEE Geoscience & Remote Sensing Letters, 2010, 7(2): 376-380.

[39] 范明虎,赵建辉,田军锋,等.星载SAR图像几何校正影响要素分析[J].雷达科学与技术,2019,17(03):246-250+256.

[40] 王少娜,刘阳,李林林,等.SAR图像变化检测综述[J].激光杂志,2023,44(01):8-15.

[41] Ammar M A, Hassan H A, Abdel-Latif M S, et al. Performance Evaluation of SAR in Presence of Multiplicative Noise Jamming[C]// 2017 34th National Radio Science Conference (NRSC). IEEE, 2017: 213-220.

[42] Aghabalaei A, Amerian Y, Ebadi H, et al. Speckle Noise Reduction of Time Series Sar Images Based on Wavelet Transform and Kalman Filter[C]// IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE, 2018: 625-628.

[43] 刘本强,赵争,盛玉婷,等.利用纹理融合与广义高斯模型的高分辨率SAR影像变化检测[J].测绘工程,2018,27(06):19-25.

[44] Monti-Guarnieri A V, Brovelli M A, Manzoni M, et al. Coherent Change Detection for Multipass SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(11): 6811-6822.

[45] 周一鸣,吴玉仁,沈项军,等.多核集成支持向量机合成孔径雷达目标分类[J].指挥信息系统与技术,2022,13(03):36-43.

[46] Son B, Kim J W, Lee D, et al. Genetic Algorithm with Species Differentiation Based on Kernel Support Vector Machine for Optimal Design of Wind Generator[J]. IEEE Transactions on Magnetics, 2019, 55(9): 1-4.

[47] Zhang Y, Peng L, Li X. et al. A Sparse Robust Adaptive Filtering Algorithm Based on the q -Rényi Kernel Function[J]. IEEE Signal Processing Letters, 2020, 27: 476-480.

[48] Yang M, Jiao L, Liu F, et al. Coarse-to-Fine Contrastive Self-Supervised Feature Learning for Land-Cover Classification in SAR Images with Limited Labeled Data[J]. IEEE Transactions on Image Processing, 2022, 31: 6502-6516.

[49] Minnehan, Breton, Savakis, et al. Grassmann Manifold Optimization for Fast $L_1$-Norm Principal Component Analysis[J]. IEEE Signal Processing Letters, 2019, 26(2): 242-246.

[50] 尹金娣.稀疏主成分分析算法研究[D].西安:西安电子科技大学,2015.

[51] Olshausen B A, Field D J. Emergence of Simple -Cell Receptive Field Properties by Learning A Sparse Coding for Natural Images[J]. Nature, 1996, 381(6583): 607-609.

[52] 董隽硕,吴玲达,郝红星.稀疏表示技术与应用综述[J].计算机系统应用, 2021,30(07):13-21.

[53] Tao L, Jiang X, Liu X, et al. Multiscale Supervised Kernel Dictionary Learning for SAR Target Recognition[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(9): 6281-6297.

[54] 张虹,左鑫兰,黄瑶.基于稀疏表示系数相关性的特征选择及SAR目标识别方法[J].激光与光电子学进展,2020,57(14):271-278.