PFWG改进的CNN多光谱遥感图像分类
|
摘要
为了实现在遥感图像处理过程中准确地提取到有效地物信息,缩短分类用时,将卷积神经网络(CNN)模型引入遥感图像地物分类,首先提出由图片模糊加权平均(PFWG)改进的CNN分类方法,利用模糊几何聚类算法作为预处理单元对实验样本进行特征规划,并对遥感地物信息进行多源特征决策,简化了分类过程,加快了CNN模型的收敛速度。实验结果表明,利用PFWG改进的CNN分类方法总体分类精度达到了93.73%;Kappa系数为0.94。该方法有效地弥补了CNN自身对遥感图像分类不够细腻、表达效果差的缺点,较好地完成了多光谱遥感图像分类任务,同时具备一定抗干扰能力。
In order to accurately achieve the effective ground information in the process of remote sensing image processing and shorten the classification time,the convolutional neural networks(CNN)model is introduced into the classification of remote sensing image features.First,the picture fuzzy weighted average(PFWG)improved CNN classification method is proposed.The fuzzy geometric clustering algorithm is used as a pre-processing unit to characterize the experimental samples,and for multi-source feature decision-making for remote sensing ground information.The classification process is simplified and the convergence of the CNN model is speeded up.The experimental results show that using PFWG improved CNN classification method,the overall classification accuracy reaches 93.73%,and the Kappa coefficient is 0.94.This method effectively compensates for the shortcoming of CNN itself which is not good enough for classification and has poor expression performance of remote sensing images.It has successfully completed an efficient classification task and has a certain anti-jamming capability.
In order to accurately achieve the effective ground information in the process of remote sensing image processing and shorten the classification time,the convolutional neural networks(CNN)model is introduced into the classification of remote sensing image features.First,the picture fuzzy weighted average(PFWG)improved CNN classification method is proposed.The fuzzy geometric clustering algorithm is used as a pre-processing unit to characterize the experimental samples,and for multi-source feature decision-making for remote sensing ground information.The classification process is simplified and the convergence of the CNN model is speeded up.The experimental results show that using PFWG improved CNN classification method,the overall classification accuracy reaches 93.73%,and the Kappa coefficient is 0.94.This method effectively compensates for the shortcoming of CNN itself which is not good enough for classification and has poor expression performance of remote sensing images.It has successfully completed an efficient classification task and has a certain anti-jamming capability.
引文
[1]Zou R X. Object-oriented high spatial resolutionremote sensing image classification using multi-feature[D]. Chengdu:University of ElectronicScience and Technology of China,2017:17-25.邹瑞雪.基于多特征的面向对象高分辨率遥感图像分类[D].成都:电子科技大学,2017:17-25.
[2]Wang Z W,Sun J J,Yu Z Y,et al.Review ofremote sensing image classification based on supportvector machine[J].Computer Science,2016,43(9):11-17,31.王振武,孙佳骏,于忠义,等.基于支持向量机的遥感图像分类研究综述[J].计算机科学,2016,43(9):11-17,31.
[3]Banerjee B,Krishna Moohan B.A novel graph basedfuzzy clustering technique for unsupervisedclassification of remote sensing images[J].ISPRSAnnals of Photogrammetry, Remote Sensing andSpatial Information Sciences,2014,II-8:165-170.
[4]Bu X B,Gong Z,Li H.Remote sensing imageclassification of improved BP neural network based onthe genetic algorithm[J]. Journal of AnhuiAgricultural Sciences,2013,41(33):13056-13058,13079.卜晓波,龚珍,黎华.基于遗传算法改进BP神经网络的遥感影像分类研究[J].安徽农业科学,2013,41(33):13056-13058,13079.
[5]Zhao S.Remote sensing image classification methodbased on convolutional neural networks[D].Beijing:China University of Geosciences,2015:14-27.赵爽.基于卷积神经网络的遥感图像分类方法研究[D].北京:中国地质大学,2015:14-27.
[6]Fu X L,Li L P,Mao K B,et al.Remote sensingimage classification based on CNN model[J].ChineseHigh Technology Letters,2017(3):203-212.付秀丽,黎玲萍,毛克彪,等.基于卷积神经网络模型的遥感图像分类[J].高技术通讯,2017(3):203-212.
[7]Liu D W,Han L,Han X Y.High spatial resolutionremote sensing image classification based on deeplearning[J].Acta Optica Sinica,2016,36(4):0428001.刘大伟,韩玲,韩晓勇.基于深度学习的高分辨率遥感影像分类研究[J].光学学报,2016,36(4):0428001.
[8]Li J,Huang X,Gamba P,et al.Multiple featurelearning for hyperspectral image classification[J].IEEE Transactions on Geoscience and RemoteSensing,2015,53(3):1592-1606.
[9]Wang M W.Image recognition by multiple classifer[D]. Beijing:Beijing University of Posts andTelecommunications,2017:11-14.王孟文.遥感图像识别分类技术研究[D].北京:北京邮电大学,2017:11-14.
[10]Wang C Y.Hesitant fuzzy set and picture fuzzy setwith their application research[D]. Changsha:Hunan University,2015:15-24.王春勇.犹豫模糊集和Picture模糊集理论与应用研究[D].长沙:湖南大学,2015:15-24.
[11]Torra V. Hesitant fuzzy sets[J].InternationalJournal of Intelligent Systems,2010,25(6):529-539.
[12]Singh P.Correlation coefficients for picture fuzzy sets[J].Journal of Intelligent &Fuzzy Systems:Applications in Engineering and Technology,2015(2):591-604.
[13]Wu Z W.Application research of convolution neuralnetwork in image classification[D]. Chengdu:University of Electronic Science and Technology,2015:12-15.吴正文.卷积神经网络在图像分类中的应用研究[D].成都:电子科技大学,2015:12-15.
[14]Simonyan K,Zisserman A.Very deep convolutionalnetworks for large-scale image recognition[J].Computer Science,2014,31(6):456-473.
[15]Han Z,Lin S Z,Zhao J C,et al.Multi-band imagefusion based on intuitionistic fuzzy set theory[J].Infrared Technology,2018,40(3):253-258.韩泽,蔺素珍,赵竞超,等.基于直觉模糊集的多波段图像融合[J].红外技术,2018,40(3):253-258.
[16]Zhang K,Hei B Q,Zhou Z,et al.CNN withcoefficient of variation-based dimensionality reductionfor hyperspectral remote sensing images classification[J].Journal of Remote Sensing,2018,22(1):87-96.张康,黑保琴,周壮,等.变异系数降维的CNN高光谱遥感图像分类[J].遥感学报,2018,22(1):87-96.
[17]Yan Q,Li H,Jing L H,et al.An extractionalgorithm of remote sensing information based onsimilarity measurement for superpixel regions[J].Laser &Optoelectronics Progress,2017,54(8):081004.闫琦,李慧,荆林海,等.一种超像素区域相似性度量的遥感信息提取算法[J].激光与光电子学进展,2017,54(8):081004.
[18]Lillesand T M,Kiefer R W,Chipman J W.Remotesensing and image interpretation[M].Peng W L,YuX C, He H,et al.Transl.Beijing:ElectronicIndustry Press,2015:366-383.利拉桑德,基弗,奇普曼.遥感与图像解译[M].彭望琭,余先川,贺辉,等译.北京:电子工业出版社,2015:366-383.
[19]Chen Y,Fan R S,Wang J X,et al.High resolutionimage classification method combining with minimumnoise fraction rotation and convolution neural network[J].Laser &Optoelectronics Progress,2017,54(10):102801.陈洋,范荣双,王竞雪,等.结合最小噪声分离变换和卷积神经网络的高分辨影像分类方法[J].激光与光电子学进展,2017,54(10):102801.
[20]Hu W, Huang Y Y, Wei L,et al. Deepconvolutional neural networks for hyperspectral imageclassification[J].Journal of Sensors,2015,2015:1-12.
[2]Wang Z W,Sun J J,Yu Z Y,et al.Review ofremote sensing image classification based on supportvector machine[J].Computer Science,2016,43(9):11-17,31.王振武,孙佳骏,于忠义,等.基于支持向量机的遥感图像分类研究综述[J].计算机科学,2016,43(9):11-17,31.
[3]Banerjee B,Krishna Moohan B.A novel graph basedfuzzy clustering technique for unsupervisedclassification of remote sensing images[J].ISPRSAnnals of Photogrammetry, Remote Sensing andSpatial Information Sciences,2014,II-8:165-170.
[4]Bu X B,Gong Z,Li H.Remote sensing imageclassification of improved BP neural network based onthe genetic algorithm[J]. Journal of AnhuiAgricultural Sciences,2013,41(33):13056-13058,13079.卜晓波,龚珍,黎华.基于遗传算法改进BP神经网络的遥感影像分类研究[J].安徽农业科学,2013,41(33):13056-13058,13079.
[5]Zhao S.Remote sensing image classification methodbased on convolutional neural networks[D].Beijing:China University of Geosciences,2015:14-27.赵爽.基于卷积神经网络的遥感图像分类方法研究[D].北京:中国地质大学,2015:14-27.
[6]Fu X L,Li L P,Mao K B,et al.Remote sensingimage classification based on CNN model[J].ChineseHigh Technology Letters,2017(3):203-212.付秀丽,黎玲萍,毛克彪,等.基于卷积神经网络模型的遥感图像分类[J].高技术通讯,2017(3):203-212.
[7]Liu D W,Han L,Han X Y.High spatial resolutionremote sensing image classification based on deeplearning[J].Acta Optica Sinica,2016,36(4):0428001.刘大伟,韩玲,韩晓勇.基于深度学习的高分辨率遥感影像分类研究[J].光学学报,2016,36(4):0428001.
[8]Li J,Huang X,Gamba P,et al.Multiple featurelearning for hyperspectral image classification[J].IEEE Transactions on Geoscience and RemoteSensing,2015,53(3):1592-1606.
[9]Wang M W.Image recognition by multiple classifer[D]. Beijing:Beijing University of Posts andTelecommunications,2017:11-14.王孟文.遥感图像识别分类技术研究[D].北京:北京邮电大学,2017:11-14.
[10]Wang C Y.Hesitant fuzzy set and picture fuzzy setwith their application research[D]. Changsha:Hunan University,2015:15-24.王春勇.犹豫模糊集和Picture模糊集理论与应用研究[D].长沙:湖南大学,2015:15-24.
[11]Torra V. Hesitant fuzzy sets[J].InternationalJournal of Intelligent Systems,2010,25(6):529-539.
[12]Singh P.Correlation coefficients for picture fuzzy sets[J].Journal of Intelligent &Fuzzy Systems:Applications in Engineering and Technology,2015(2):591-604.
[13]Wu Z W.Application research of convolution neuralnetwork in image classification[D]. Chengdu:University of Electronic Science and Technology,2015:12-15.吴正文.卷积神经网络在图像分类中的应用研究[D].成都:电子科技大学,2015:12-15.
[14]Simonyan K,Zisserman A.Very deep convolutionalnetworks for large-scale image recognition[J].Computer Science,2014,31(6):456-473.
[15]Han Z,Lin S Z,Zhao J C,et al.Multi-band imagefusion based on intuitionistic fuzzy set theory[J].Infrared Technology,2018,40(3):253-258.韩泽,蔺素珍,赵竞超,等.基于直觉模糊集的多波段图像融合[J].红外技术,2018,40(3):253-258.
[16]Zhang K,Hei B Q,Zhou Z,et al.CNN withcoefficient of variation-based dimensionality reductionfor hyperspectral remote sensing images classification[J].Journal of Remote Sensing,2018,22(1):87-96.张康,黑保琴,周壮,等.变异系数降维的CNN高光谱遥感图像分类[J].遥感学报,2018,22(1):87-96.
[17]Yan Q,Li H,Jing L H,et al.An extractionalgorithm of remote sensing information based onsimilarity measurement for superpixel regions[J].Laser &Optoelectronics Progress,2017,54(8):081004.闫琦,李慧,荆林海,等.一种超像素区域相似性度量的遥感信息提取算法[J].激光与光电子学进展,2017,54(8):081004.
[18]Lillesand T M,Kiefer R W,Chipman J W.Remotesensing and image interpretation[M].Peng W L,YuX C, He H,et al.Transl.Beijing:ElectronicIndustry Press,2015:366-383.利拉桑德,基弗,奇普曼.遥感与图像解译[M].彭望琭,余先川,贺辉,等译.北京:电子工业出版社,2015:366-383.
[19]Chen Y,Fan R S,Wang J X,et al.High resolutionimage classification method combining with minimumnoise fraction rotation and convolution neural network[J].Laser &Optoelectronics Progress,2017,54(10):102801.陈洋,范荣双,王竞雪,等.结合最小噪声分离变换和卷积神经网络的高分辨影像分类方法[J].激光与光电子学进展,2017,54(10):102801.
[20]Hu W, Huang Y Y, Wei L,et al. Deepconvolutional neural networks for hyperspectral imageclassification[J].Journal of Sensors,2015,2015:1-12.