模糊区域级MRF方法在城镇自动识别中的应用
|
摘要
针对遥感影像中城镇区域内外的自然地物难以区分,城镇区域不易完整识别的问题,提出一种对象级模糊MRF识别方法.该方法首先通过光谱信息和空间梯度分析得到城镇种子点(人造地物顶部点和阴影点);然后由均值漂移算法过分割影像;再对过分割区域建立MRF,在迭代过程中用MRF的条件概率矩阵代替模糊C均值聚类算法的隶属度矩阵,并保持包含种子点的区域类别不变,从而实现城镇识别.对于Quick Bird和Ikonos遥感影像,该模型能够兼顾城镇区域自动识别过程中的随机性与模糊性,很好地利用了空间相关信息,有效识别出了城镇区域.
In order to effectively distinguish the natural objects inside the town from outside objects, and completely identify the urban regions in remote sensing image,a fuzzy geographic object-based MRF method is proposed. Firstly, the seed points of the town,i. e., the top and shadow points of artificial ground objects, are firstly obtained by analyzing spectral information and spatial gradient. Then the over-segmented regions of the origin image are obtained by Mean Shift algorithm.Finally,a MRF is established over regions. and the membership matrix in the fuzzy C-means clustering algorithm is replaced by the conditional probability matrix in the MRF in an iterative manner. Meanwhile the categories of the regions containing the seed points are kept unchanged. For QuickBird and Ikonos remote sensing images, the proposed model can simultaneously deal with both the stochastic and fuzzy nature of images, and effectively intergate the space information, and thus benefit the identification of town.
In order to effectively distinguish the natural objects inside the town from outside objects, and completely identify the urban regions in remote sensing image,a fuzzy geographic object-based MRF method is proposed. Firstly, the seed points of the town,i. e., the top and shadow points of artificial ground objects, are firstly obtained by analyzing spectral information and spatial gradient. Then the over-segmented regions of the origin image are obtained by Mean Shift algorithm.Finally,a MRF is established over regions. and the membership matrix in the fuzzy C-means clustering algorithm is replaced by the conditional probability matrix in the MRF in an iterative manner. Meanwhile the categories of the regions containing the seed points are kept unchanged. For QuickBird and Ikonos remote sensing images, the proposed model can simultaneously deal with both the stochastic and fuzzy nature of images, and effectively intergate the space information, and thus benefit the identification of town.
引文
[1]Xueliang Zhang,Pengfeng Xiao,Xuezhi Feng,Min Yuan.Separate segmentation of multi-temporal high-resolution re-mote sensing images for object-based change detection inurban area[J]. Remote Sensing of Environment,2017,201:243-255.
[2]Michael Meinild Nielsen. Extraction of different urban areacategories from satellite images using Window IndependentContext Segmentation[A]. Conference 2011 Joint UrbanRemote Sensing Event[C]. M unich Germany,IEEE,April2011. 11-13
[3]Essa Basaeed,Harish Bhaskar,Mohammed Al-Mualla. Su-pervised remote sensing image segmentation using boostedconvolutional neural netw orks[J]. Know ledge-Based Sys-tems,2016,99:19-27.
[4] Yi Ding,Xian Fu. Kernel-based fuzzy c-means clusteringalgorithm based on genetic algorithm[J]. Neurocomput-ing,2016,188:233-238.
[5]赵泉华,李晓丽,赵雪梅,李玉.结合马氏距离的区域化模糊聚类遥感图像分割[J].中国矿业大学学报,2017,46(1):222-228.Zhao Quanhua,Li Xiaoli,Zhao Xuemei,Li Yu. Remotesensing image segmentation algorithm w ith regional fuzzycluster and M ahalanobis distance[J]. Journal of China Uni-versity of M ining&Technology,2017,46(1):222-228.(in Chinese)
[6] Kun Li,Ai Min Li. Modified FCM image segmentationbased on Gaussian kernel and spatial information[A].Conference 2017 IEEE 2nd Information Technology,Net-w orking,Electronic and Automation Control Conference[C]. Chongqing China,IEEE,2017. 1105-1109.
[7]陈荣元,徐雪松,李广琼,申立智,刘星宝.自适应特征加权的Gibbs随机场影像分割方法[J].电子学报,2016,44(10):2351-2356.CHEN Rong-yuan,XU Xue-song,et al. Image segmenta-tion by combining adaptively w eighted features w ith gibbsrandom field[J]. Acta Electronica Sinica,2016,44(10):2351-2356.(in Chinese)
[8]G-S Xia,C He,H Sun. Integration of synthetic aperture ra-dar image segmentation method using M arkov random fieldon region adjacency graph[J]. IET Radar Sonar&Naviga-tion,2007,1(5):348-353.
[9]Xiao Feng Wang,Xiao-Ping Zhang. A new localized super-pixel M arkov random field for image segmentation[A].Conference 2009 IEEE international conference on M ulti-media and Expo[C]. New York America,IEEE,2009. 642-645.
[10] Guoying Liu,Qianqing Qin,Tiancan Mei,LieguangWang. Supervised image segmentation based on tree-structured M RF model in w avelet domain[J]. IEEE Ge-oscience and Remote Sensing Letters. 2009,6(4):850-854.
[11] Chen Zheng,Leiguang Wang. Semantic Segmentation ofRemote Sensing Imagery Using Object-Based M arkovRandom Field M odel With Regional Penalties[J]. IEEEJournal of Selected Topics in Applied Earth Observationsand Remote Sensing,2015,8(5):1924-1935.
[12]John M. Hammersley,Peter Clifford. Markov fields on finitegraphs and lattices[M]. Unpublished manuscript,1971.
[13] Yong Xia,Dagan Feng,Rongchun Zhao. Adaptive Seg-mentation of Textured Images by Using the CoupledM arkov Random Field M odel[J]. IEEE Transactions onImage Processing,2006,15(11):3559-3566.
[14]Sadaaki Miyamoto,Masao Mukaidono. Fuzzy c-means asa regularization and maximum entropy approach[A].Conference 7th International Fuzzy Systems AssociationWorld Congress[C]. Prague,Czech Republic,FIZKarlsruhe GmbH. 1997. 2:86-92.
[15] Hidetomo Ichihashi,Kiyotaka Miyagishi,Katsubiro Hon-da. Fuzzy c-means clustering w ith regularization by K-Linformation[A]. Conference The 10th IEEE InternationalConference on Fuzzy Systems[C]. M elbourne,IEEE,2001. 2:924-927.
[16] Sotirios P. Chatzis,Theodora A. Varvarigou. A FuzzyClustering Approach Tow ard Hidden M arkov RandomField M odels for Enhanced Spatially Constrained ImageSegmentation[J]. IEEE Transactions on Fuzzy Systems,2008,16(5):1351-1361.
[17]陈荣元,郑晨,王雷光,秦前清. MRF框架下的区域增长模型在城镇识别中的应用[J].测绘学报,2011,40(2):163-168.Chen Rongyuan,Zheng Chen,Wan Leiguang,Qin Qian-qing. A region grow ing model under the framew ork ofM RF for urban detection[J]. Acta Geoda etica et Ca rto-graphica Sinica,2011,40(2):163-168.(in Chinese)
[2]Michael Meinild Nielsen. Extraction of different urban areacategories from satellite images using Window IndependentContext Segmentation[A]. Conference 2011 Joint UrbanRemote Sensing Event[C]. M unich Germany,IEEE,April2011. 11-13
[3]Essa Basaeed,Harish Bhaskar,Mohammed Al-Mualla. Su-pervised remote sensing image segmentation using boostedconvolutional neural netw orks[J]. Know ledge-Based Sys-tems,2016,99:19-27.
[4] Yi Ding,Xian Fu. Kernel-based fuzzy c-means clusteringalgorithm based on genetic algorithm[J]. Neurocomput-ing,2016,188:233-238.
[5]赵泉华,李晓丽,赵雪梅,李玉.结合马氏距离的区域化模糊聚类遥感图像分割[J].中国矿业大学学报,2017,46(1):222-228.Zhao Quanhua,Li Xiaoli,Zhao Xuemei,Li Yu. Remotesensing image segmentation algorithm w ith regional fuzzycluster and M ahalanobis distance[J]. Journal of China Uni-versity of M ining&Technology,2017,46(1):222-228.(in Chinese)
[6] Kun Li,Ai Min Li. Modified FCM image segmentationbased on Gaussian kernel and spatial information[A].Conference 2017 IEEE 2nd Information Technology,Net-w orking,Electronic and Automation Control Conference[C]. Chongqing China,IEEE,2017. 1105-1109.
[7]陈荣元,徐雪松,李广琼,申立智,刘星宝.自适应特征加权的Gibbs随机场影像分割方法[J].电子学报,2016,44(10):2351-2356.CHEN Rong-yuan,XU Xue-song,et al. Image segmenta-tion by combining adaptively w eighted features w ith gibbsrandom field[J]. Acta Electronica Sinica,2016,44(10):2351-2356.(in Chinese)
[8]G-S Xia,C He,H Sun. Integration of synthetic aperture ra-dar image segmentation method using M arkov random fieldon region adjacency graph[J]. IET Radar Sonar&Naviga-tion,2007,1(5):348-353.
[9]Xiao Feng Wang,Xiao-Ping Zhang. A new localized super-pixel M arkov random field for image segmentation[A].Conference 2009 IEEE international conference on M ulti-media and Expo[C]. New York America,IEEE,2009. 642-645.
[10] Guoying Liu,Qianqing Qin,Tiancan Mei,LieguangWang. Supervised image segmentation based on tree-structured M RF model in w avelet domain[J]. IEEE Ge-oscience and Remote Sensing Letters. 2009,6(4):850-854.
[11] Chen Zheng,Leiguang Wang. Semantic Segmentation ofRemote Sensing Imagery Using Object-Based M arkovRandom Field M odel With Regional Penalties[J]. IEEEJournal of Selected Topics in Applied Earth Observationsand Remote Sensing,2015,8(5):1924-1935.
[12]John M. Hammersley,Peter Clifford. Markov fields on finitegraphs and lattices[M]. Unpublished manuscript,1971.
[13] Yong Xia,Dagan Feng,Rongchun Zhao. Adaptive Seg-mentation of Textured Images by Using the CoupledM arkov Random Field M odel[J]. IEEE Transactions onImage Processing,2006,15(11):3559-3566.
[14]Sadaaki Miyamoto,Masao Mukaidono. Fuzzy c-means asa regularization and maximum entropy approach[A].Conference 7th International Fuzzy Systems AssociationWorld Congress[C]. Prague,Czech Republic,FIZKarlsruhe GmbH. 1997. 2:86-92.
[15] Hidetomo Ichihashi,Kiyotaka Miyagishi,Katsubiro Hon-da. Fuzzy c-means clustering w ith regularization by K-Linformation[A]. Conference The 10th IEEE InternationalConference on Fuzzy Systems[C]. M elbourne,IEEE,2001. 2:924-927.
[16] Sotirios P. Chatzis,Theodora A. Varvarigou. A FuzzyClustering Approach Tow ard Hidden M arkov RandomField M odels for Enhanced Spatially Constrained ImageSegmentation[J]. IEEE Transactions on Fuzzy Systems,2008,16(5):1351-1361.
[17]陈荣元,郑晨,王雷光,秦前清. MRF框架下的区域增长模型在城镇识别中的应用[J].测绘学报,2011,40(2):163-168.Chen Rongyuan,Zheng Chen,Wan Leiguang,Qin Qian-qing. A region grow ing model under the framew ork ofM RF for urban detection[J]. Acta Geoda etica et Ca rto-graphica Sinica,2011,40(2):163-168.(in Chinese)