基于GOF的SAR海冰图像无监督自动分类算法
|
摘要
提出一种基于拟合优度(GOF)的合成孔径雷达(SAR)海冰图像无监督自动分类算法。该算法首先在高斯混合模型(GMM)参数进行估计过程中,附加了一个拟合优度(GOF)检验过程,在一定置信区间条件下,不仅能动态地选择最佳显著类别数,而且在马尔科夫随机场(MRF)计算最小能量对区域重新标记时,能提供初始特征参数;然后,再进行区域增长循环迭代;最后,得到各区域的最佳标记。实验结果证明,此算法在提高运行效率的同时,能够解决无监督图像分类过程中需要手动输入类别数的问题,而且输出结果图与专家解译的实况图相比在细节保持上效果更好,实用性更强。
An unsupervised automatic classification algorithm for sea ice image of SAR based on goodness-of-fit test( GOF),which named GOF MRF is proposed. Firstly,in the process of estimating Gaussian mixture mode( GMM) parameters,a GOF process is added,which dynamically selects the best number of significant classes and provides an initial feature parameter to calculate the MRF minimum energy and remark the region under certain confidence intervals. Since the initial feature parameter selection of the MRF is not random,the operation efficiency is also improved while reducing the number of iteration cycles of the algorithm. The experiment results demonstrate that this algorithm solves the problem of manual input of the number of classes in the unsupervised image classification process,the output result graph is also better in terms of detail maintenance than the expert interpretation of the truth map,and can support the operation and meet the real-time requirements.
An unsupervised automatic classification algorithm for sea ice image of SAR based on goodness-of-fit test( GOF),which named GOF MRF is proposed. Firstly,in the process of estimating Gaussian mixture mode( GMM) parameters,a GOF process is added,which dynamically selects the best number of significant classes and provides an initial feature parameter to calculate the MRF minimum energy and remark the region under certain confidence intervals. Since the initial feature parameter selection of the MRF is not random,the operation efficiency is also improved while reducing the number of iteration cycles of the algorithm. The experiment results demonstrate that this algorithm solves the problem of manual input of the number of classes in the unsupervised image classification process,the output result graph is also better in terms of detail maintenance than the expert interpretation of the truth map,and can support the operation and meet the real-time requirements.
引文
[1]LANG W H,YANG S,ZHANG X.Segmentation of SARsea ice image based on region splitting and adaptive refinement process[J].Journal of Remote Sensing,2015,19(5):864-872.
[2]SANTORO M.Retrieval of growing stock volume in boreal forest using hyper-temporal series of Envisat ASARScan SAR backscatter measurements[J]Remote Sensing of Environment,2011,115(2):490-507.
[3]KWOK R,RIGNOT E,HOLT B,et al.Identification of sea ice types in spaceborne synthetic aperture radar data[J].Journal of Geophysical Research:Oceans,2012,97(52):2391-2402.
[4]OCHILOV S,CLAUSI D A.Operational SAR sea-ice image classification[J].IEEE Transactions on Geoscience&Remote Sensing,2012,50(11):4397-4408.
[5]OCHILOV S,CLAUSI D A.Automated Classification of Operational SAR Sea Ice Images[C].Computer and Robot Vision,2010:40-46.
[6]LI F,CLAUSI D A,WANG L,et al.A semi-supervised approach for ice-water classification using dualpolarization SAR satellite imagery[C].Computer Vision and Pattern Recognition,2015:28-35.
[7]LEIGH S,WANG Z,CLAUSI D A.Automated ice:Water classification using dual polarization SAR satellite imagery[J].IEEE Transactions on Geoscience&Remote Sensing,2013,52(9):5529-5539.
[8]ZAKHVATJINA N Y,ALEXANDROV V Y,JOHANNESSEN O M,et al.Classification of sea ice types in ENVISAT synthetic aperture radar images[J].IEEE Transactions on Geoscience&Remote Sensing,2013,51(5):2587-2600.
[9]LI C T.A conditional random field approach to unsupervised texture image segmentation[J].Eurasip Journal on Advances in Signal Processing,2010,16(1):16-22.
[10]LANG W,ZHANG P,Wu J,et al.Incidence angle correction of SAR sea ice data based on locally linear mapping[J].IEEE Transactions on Geoscience&Remote Sensing,2016,54(6):3188-3199.
[11]SALVADOR S,CHAN P.Determining the number of clusters/segments in hierarchical clustering/segmentation algorithms[C].IEEE International Conference on Tools with Artificial Intelligence,2008:576-584.
[12]CAO F,HONG W,WU Y,et al.An Unsupervised segmentation with an adaptive number of clusters using the SPAN/H/alpha/A space and the complex wishart clustering for fully polarimetric SAR data analysis[J].IEEE Transactions on Geoscience and Remote Sensing,2007,45(11):3454-3467.
[13]MEMARSADEGHI N,MOUNT D.M,NETANYAHU NS,et al.A fast implementation of the ISODATAclustering algorithm[J].International Journal of Computational Geometry and Applications,2007,17(10):71-103.
[14]DOUGERIS A P,ANFINSEN S N,ELTOFT T.Automated non-gaussian clustering of polarimetric synthetic aperture radar images[J].IEEE Transactions on Geoscience&Remote Sensing,2011,49(10):3665-3676.
[15]DOUGERIS A.ELTOFT T.Scale mixture of Gaussians modelling of polarimetric SAR data[C].International POL in SAR Workshop,2007:22-26.
[16]YU Q.Automated SAR sea ice interpretation[D].Waterloo:University of Waterloo,2006.
[17]LIU Z,WU Z,LI T,et al.GMM and CNN hybrid method for short utterance speaker recognition[J].IEEETransactions on Industrial Informatics,2018,56(10):123-132.
[18]STEVEN L.Automated SAR sea ice interpretation[D].Waterloo:University of Waterloo,2006.
[19]STEVEN L.Automated ice:Water classification using dual polarization SAR satellite imagery[D].Waterloo:University of Waterloo,2013.
[20]YANG S,LANG W H,Jie Wu,Xuezhi Yang.Combining MRF andν-SVM for SAR sea ice image classification[J].Journal of Remote Sensing,2015,19(5):844-855.
[21]YU Q,CLAUSI D A.IRGS:Image segmentation using edge penalties and region growing[J].IEEE Trans on Pattern Anal.Mach.2008,30(12):2126-2139.
[22]WANG L,SCOTT K A,XU L,et al.Sea ice concentration estimation during melt from dual-pol SARscenes using deep convolutional neural networks:A case study[J].IEEE Transactions on Geoscience&Remote Sensing,2016,54(8):4524-4533.
[2]SANTORO M.Retrieval of growing stock volume in boreal forest using hyper-temporal series of Envisat ASARScan SAR backscatter measurements[J]Remote Sensing of Environment,2011,115(2):490-507.
[3]KWOK R,RIGNOT E,HOLT B,et al.Identification of sea ice types in spaceborne synthetic aperture radar data[J].Journal of Geophysical Research:Oceans,2012,97(52):2391-2402.
[4]OCHILOV S,CLAUSI D A.Operational SAR sea-ice image classification[J].IEEE Transactions on Geoscience&Remote Sensing,2012,50(11):4397-4408.
[5]OCHILOV S,CLAUSI D A.Automated Classification of Operational SAR Sea Ice Images[C].Computer and Robot Vision,2010:40-46.
[6]LI F,CLAUSI D A,WANG L,et al.A semi-supervised approach for ice-water classification using dualpolarization SAR satellite imagery[C].Computer Vision and Pattern Recognition,2015:28-35.
[7]LEIGH S,WANG Z,CLAUSI D A.Automated ice:Water classification using dual polarization SAR satellite imagery[J].IEEE Transactions on Geoscience&Remote Sensing,2013,52(9):5529-5539.
[8]ZAKHVATJINA N Y,ALEXANDROV V Y,JOHANNESSEN O M,et al.Classification of sea ice types in ENVISAT synthetic aperture radar images[J].IEEE Transactions on Geoscience&Remote Sensing,2013,51(5):2587-2600.
[9]LI C T.A conditional random field approach to unsupervised texture image segmentation[J].Eurasip Journal on Advances in Signal Processing,2010,16(1):16-22.
[10]LANG W,ZHANG P,Wu J,et al.Incidence angle correction of SAR sea ice data based on locally linear mapping[J].IEEE Transactions on Geoscience&Remote Sensing,2016,54(6):3188-3199.
[11]SALVADOR S,CHAN P.Determining the number of clusters/segments in hierarchical clustering/segmentation algorithms[C].IEEE International Conference on Tools with Artificial Intelligence,2008:576-584.
[12]CAO F,HONG W,WU Y,et al.An Unsupervised segmentation with an adaptive number of clusters using the SPAN/H/alpha/A space and the complex wishart clustering for fully polarimetric SAR data analysis[J].IEEE Transactions on Geoscience and Remote Sensing,2007,45(11):3454-3467.
[13]MEMARSADEGHI N,MOUNT D.M,NETANYAHU NS,et al.A fast implementation of the ISODATAclustering algorithm[J].International Journal of Computational Geometry and Applications,2007,17(10):71-103.
[14]DOUGERIS A P,ANFINSEN S N,ELTOFT T.Automated non-gaussian clustering of polarimetric synthetic aperture radar images[J].IEEE Transactions on Geoscience&Remote Sensing,2011,49(10):3665-3676.
[15]DOUGERIS A.ELTOFT T.Scale mixture of Gaussians modelling of polarimetric SAR data[C].International POL in SAR Workshop,2007:22-26.
[16]YU Q.Automated SAR sea ice interpretation[D].Waterloo:University of Waterloo,2006.
[17]LIU Z,WU Z,LI T,et al.GMM and CNN hybrid method for short utterance speaker recognition[J].IEEETransactions on Industrial Informatics,2018,56(10):123-132.
[18]STEVEN L.Automated SAR sea ice interpretation[D].Waterloo:University of Waterloo,2006.
[19]STEVEN L.Automated ice:Water classification using dual polarization SAR satellite imagery[D].Waterloo:University of Waterloo,2013.
[20]YANG S,LANG W H,Jie Wu,Xuezhi Yang.Combining MRF andν-SVM for SAR sea ice image classification[J].Journal of Remote Sensing,2015,19(5):844-855.
[21]YU Q,CLAUSI D A.IRGS:Image segmentation using edge penalties and region growing[J].IEEE Trans on Pattern Anal.Mach.2008,30(12):2126-2139.
[22]WANG L,SCOTT K A,XU L,et al.Sea ice concentration estimation during melt from dual-pol SARscenes using deep convolutional neural networks:A case study[J].IEEE Transactions on Geoscience&Remote Sensing,2016,54(8):4524-4533.