吉林大学学报(工学版) ›› 2015, Vol. 45 ›› Issue (4): 1353-1359.doi: 10.13229/j.cnki.jdxbgxb201504047

• • 上一篇    下一篇

基于MSA特征和模拟退火优化的遥感图像多目标关联算法

李晖晖1, 滑立1, 杨宁1, 刘坤2   

  1. 1.西北工业大学 自动化学院,西安 710072;
    2.上海海事大学 信息工程学院,上海 200135
  • 收稿日期:2013-12-01 出版日期:2015-07-01 发布日期:2015-07-01
  • 作者简介:李晖晖(1974-),女,副教授,博士.研究方向:数字图像处理,图像信息融合.E-mail:lihhui@nwpu.edu.cn
  • 基金资助:
    航空基金(20131953022); 西北工业大学基础研究基金项目(JC20110266); 装备研究基金项目(9140A06050113HK****)

Multi-target association algorithm for remote sensing images based on MSA features and simulated annealing optimization

LI Hui-hui1, HUA Li1, YANG Ning1, LIU Kun2   

  1. 1.College of Automation, Northwestern Polytechnical University, Xi'an 710072,China;
    2.School of Information Engineering, Shanghai Maritime University, Shanghai 200135, China
  • Received:2013-12-01 Online:2015-07-01 Published:2015-07-01

摘要: 由于当前遥感成像技术一般只能获取采样稀疏的遥感图像,无法准确估计目标的状态信息,因此传统的利用状态特征进行关联的方法并不适合遥感图像的目标关联。选取不依赖于时间的目标图像特征作为关联量又无法处理大场景中多个目标关联引起的模糊性。针对上述问题,本文提出了基于多尺度自卷积不变矩特征匹配和模拟退火优化的多目标关联算法。首先提取目标的多尺度自卷积矩(MSA)特征,计算特征间匹配概率,构造整体关联代价矩阵,并设置自适应温度更新函数和双阈值对模拟退火算法进行改进,快速寻求全局最优解。实验结果表明,该算法能够有效地利用遥感图像特征信息,消除关联模糊性,高效解决多目标关联问题。

关键词: 摄影测量与遥感技术, 目标关联, MSA特征, 关联代价矩阵, 模拟退火算法

Abstract: Since the current remote sensing imaging techniques can only get sparse sampling remote images, which are unable to accurately estimate the target state information, so the traditional association method using state features is not suitable for remote sensing image target. If the time independent image features are selected as the associative volume, the ambiguity induced by the association multiple targets in big scene can not be handled. To solve the above problem, a multi-scale autoconvolution feature matching and simulated annealing optimization algorithm is proposed. First, the MSA features are extracted and the matching probability between features is calculated to construct the whole association cost matrix. Then, the adaptive temperature updating function and the double threshold are set to improve the simulated annealing algorithm for quick search of the global optimal solution. Experimental results show that the proposed algorithm can effectively use the feature information of remote sensing image, eliminate association ambiguity and solve the problem of multi-target association efficiently.

Key words: photogrammetry and remote sensing, target association, msa feature, association cost matrix, simulated annealing algorithm

中图分类号: 

  • TP751
[1] 雷琳.多源遥感图像舰船目标特征提取和融合技术研究[D].长沙:国防科技大学电子科学与工程学院,2008. Lei Lin. Ship feature extraction and fusion in multiple remote sensing images[D]. Changsha: School of Electronic Science and Engineering,National University of Defense Technology, 2008.
[2] Bar-shalom Y, Sherlukde H M, Pattipati K R. Use of measurements from an image sensor for precision target tracking[J]. IEEE Trans AES, 1989, 25(6):863-871.
[3] Bar Shalom Y. Extension of the probabilistic data association filter in multi-target tracking[C]∥Proceedings of the 5th symposium on nonlinear estimation, 1974:16-21.
[4] Blackman S, Dempster R J, Broida T J. Multiple hypothesis track confirmation for infrared surveillance systems[J]. IEEE Trans AES, 1993, 29(3):810-823.
[5] 董学志, 宋建中, 韩广良.一种利用Gabor小波特征的目标跟踪方法[J].光学技术,2003,29(4):484-486. Dong Xue-zhi, Song Jian-zhong, Han Guang-liang. Target tracking method using wavelet feature[J]. Optical Technique, 2003,29(4):484-486.
[6] 姚剑, 刘其真, 张斌.模糊技术与神经网络的混合算法在运动目标识别与跟踪中的应用[J].计算机工程与应用, 2000,1:62-64. Yao Jian, Liu Qi-zhen, Zhang Bin. The application of combination of fuzzy algorithms and neural network in tracking moving object[J].Computer Engineering and Applications, 2000,1:62-64.
[7] Hu M K. Pattern recognition by moment invariant[C]∥Proc IRE, 1961, 49: 1428-1436.
[8] Rahtu E,Salo M,Heikkila J.Affine invariant pattern recognition using multi-scale auto convolution[J].IEEE Transactions on Pattern analysis and Machine Intelligence,2005,27(6):908-918.
[9] 雷琳, 蔡红苹, 唐涛, 等.基于MSA特征的遥感图像多目标关联算法[J].遥感学报,2008, 12(4):586-592. Lei Lin, Cai Hong-ping, Tang Tao, et al. A MSA feature-based multiple targets association algorithm in remote sensing images[J]. Journal of Remote Sensing, 2008, 12(4):586-592.
[10] Bar-Shalom Y, Kirubarajan T, Gokberk C. Tracking with classification aided multiframe data assoeiation[J]. IEEE Transactions on Aerospace and Electronic Systems, 2005, 41(3):868-878.
[11] Mandal A K, Pal S, De A K, et al. Novel approach to identify good tracer clouds from a sequence of satellite images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(4):813-818.
[12] Krickpatric S, Gelett J C D, Vecchi M P. Optimization by simulated annealing[J]. Science, 1983, 220(4598): 671-680.
[13] Winkler G. Image Analysis,Random Fields and Dynamic Monte Carlo Methods[M]. Berlin: Springer-Verlag, 1999.
[14] 高尚.模拟退火算法中的退火策略研究[J].航空计算技术, 2002,32(4):20-26. Gao Shang. Research on annealing strategy in simulated annealing algorithm[J]. Aeronautical Computer Technique, 2002,32(4):20-26.
[1] 代存杰,李引珍,马昌喜,柴获,牟海波. 不确定条件下危险品配送路线多准则优化[J]. 吉林大学学报(工学版), 2018, 48(6): 1694-1702.
[2] 姜超, 耿则勋, 刘立勇, 潘映峰. 结合噪声去除的极大似然图像复原[J]. 吉林大学学报(工学版), 2015, 45(4): 1360-1366.
[3] 王京萌, 张爱武, 赵宁宁, 孟宪刚. 斜采样的倾斜角度对采样产生混叠的影响及其与分辨率的关系[J]. 吉林大学学报(工学版), 2015, 45(3): 953-960.
[4] 韩啸,刘淑芬,徐天琦. 基于遗传模拟退火算法的改进K-medoids算法[J]. 吉林大学学报(工学版), 2015, 45(2): 619-623.
[5] 曹建农, 郭佳, 王蓓, 董昱威, 王平禄. 高分辨率影像中城区树冠多尺度聚类识别方法[J]. 吉林大学学报(工学版), 2014, 44(4): 1215-1224.
[6] 刘逻, 郭立红, 肖辉, 王建军, 王改革. 基于结合模拟退火算法的动态模糊神经网络的软件可靠性增长模型[J]. , 2012, 42(05): 1225-1230.
[7] 李玉庆,徐敏强,王日新 . 三轴稳定卫星点目标观测任务优化调度技术[J]. 吉林大学学报(工学版), 2008, 38(06): 1447-1451.
[8] 李宝林;李志蜀;金虎;孙继荣;陈艳红 . 基于R_N(K)准则的测试用例生成技术[J]. 吉林大学学报(工学版), 2008, 38(03): 680-0684.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!