基于卷积网络的沙漠腹地绿洲植物群落自动分类方法
|
摘要
为解决沙漠腹地绿洲遥感图像植物群落背景较易混淆,仅用传统的基于像元光谱信息的图像处理方法未能充分利用其图像特征信息,使得提取效果不佳的问题,针对地物类内特征复杂、类间边界模糊的特点,以连续分布的区域为研究对象,提出了一种基于深度卷积神经网络(Convolutional neural network,CNN)的高分辨率遥感影像植物群落自动分类方法。切分无人机影像获得规则块图像,利用基于CNN的VGGNet和Res Net模型分别对块图像的特征进行抽象与学习,以自动获取更加深层抽象、更具代表性的图像块深层特征,从而实现对植物群落分布区域的提取,以原图像与结果图像叠加的形式输出植物群落自动分类结果。采用了不同梯度的样本数量作为训练样本,利用文中提出的方法分析了不同梯度的训练样本数量对自动分类结果的影响。实验结果表明,训练样本数量对分类精度具有明显的影响;提高其泛化能力后,Res Net50模型与VGG19模型的建模精度从86. 00%、83. 33%分别提升到92. 56%、90. 29%; Res Net50模型分类精度为83. 53%～91. 83%,而VGG19模型分类精度为80. 97%～89. 56%,与传统的监督分类方法比较,深度卷积网络明显提高了分类精度。分类结果表明,训练样本数量不低于200时,基于CNN的Res Net50模型表现出最佳的分类结果。
In order to solve the problem of remote sensing image plant community background,only the traditional image processing method based on pixel spectral information fails to make full use of its image feature information,which makes the extraction effect poor. Aiming at the complex features of plant species and the blurring of inter-class boundaries,the continuous distribution of regions was taken as the research object. A high-resolution remote sensing image plant community automatic classification based on the convolutional neural network( CNN) was proposed. The UAV images were segmented to obtain regular block images,and the features of block images were abstracted and learned by CNN-based VGGNet and ResNet models to automatically acquire deeper abstract and more representative image block deep features. The extraction of the plant community distribution area was performed to output the automatic classification results of the plant community in the form of superposition of the original image and the result image. The number of samples with different gradients was used as the training sample.The influence of the number of training samples with different gradients on the automatic classification results was analyzed by the proposed method. The experimental results showed that the number of training samples had a significant impact on the classification accuracy. After improving its generalization ability,the modeling accuracy of ResNet50 model and VGG19 model was improved from 86. 00% and 83. 33% to92. 56% and 90. 29%,respectively. The classification accuracy of ResNet50 model was varied from83. 53% to 91. 83%,while the classification accuracy of the VGG19 model was varied from 80. 97% to89. 56%. Compared with the traditional supervised classification method,the deep convolution network significantly improved the classification accuracy. Through the analysis of classification result,it was found that the number of training samples should not be less than 200,and the CNN-based ResNet50 model showed the best classification results.
In order to solve the problem of remote sensing image plant community background,only the traditional image processing method based on pixel spectral information fails to make full use of its image feature information,which makes the extraction effect poor. Aiming at the complex features of plant species and the blurring of inter-class boundaries,the continuous distribution of regions was taken as the research object. A high-resolution remote sensing image plant community automatic classification based on the convolutional neural network( CNN) was proposed. The UAV images were segmented to obtain regular block images,and the features of block images were abstracted and learned by CNN-based VGGNet and ResNet models to automatically acquire deeper abstract and more representative image block deep features. The extraction of the plant community distribution area was performed to output the automatic classification results of the plant community in the form of superposition of the original image and the result image. The number of samples with different gradients was used as the training sample.The influence of the number of training samples with different gradients on the automatic classification results was analyzed by the proposed method. The experimental results showed that the number of training samples had a significant impact on the classification accuracy. After improving its generalization ability,the modeling accuracy of ResNet50 model and VGG19 model was improved from 86. 00% and 83. 33% to92. 56% and 90. 29%,respectively. The classification accuracy of ResNet50 model was varied from83. 53% to 91. 83%,while the classification accuracy of the VGG19 model was varied from 80. 97% to89. 56%. Compared with the traditional supervised classification method,the deep convolution network significantly improved the classification accuracy. Through the analysis of classification result,it was found that the number of training samples should not be less than 200,and the CNN-based ResNet50 model showed the best classification results.
引文
[1]王倩.北京植物园植物群落案例分析[J].福建农业,2015(4):235.
[2]张同瑞,赵庚星,高明秀,等.基于近地多光谱和OLI影像的黄河三角洲冬小麦种植区盐分估算及遥感反演——以山东省垦利县和无棣县为例[J].自然资源学报,2016,31(6):1051-1060.ZHANG Tongrui,ZHAO Gengxing,GAO Mingxiu,et al. Soil salinity estimation and remote sensing inversion based on nearground multispectral and TM imagery in winter wheat growing area in the Yellow River Delta—case study in Kenli County and Wudi County,Shandong Province[J]. Journal of Natural Resources,2016,31(6):1051-1060.(in Chinese)
[3]肖海燕,曾辉,昝启杰,等.基于高光谱数据和专家决策法提取红树林群落类型信息[J].遥感学报,2007,11(4):531-537.XIAO Haiyan,ZENG Hui,ZAN Qijie,et al. Decision tree model in extraction of mangrove community information using hyperspectral image data[J]. Journal of Remote Sensing,2007,11(4):531-537.(in Chinese)
[4]杜欣,黄晓霞,李红旮,等.基于投影寻踪学习网络算法的植物群落高分遥感分类研究[J].地球信息科学学报,2016,18(1):124-132.DU Xin,HUANG Xiaoxia,LI Hongga,et al. Research on classification of plant community using projection pursuit learning network algorithm on high resolution remote sensing images[J]. Journal of Geo-information Science,2016,18(1):124-132.(in Chinese)
[5]鲁恒,付萧,贺一楠,等.基于迁移学习的无人机影像耕地信息提取方法[J/OL].农业机械学报,2015,46(12):274-279,284. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? file_no=20151237&flag=1. DOI:10. 6041/j. issn.1000-1298. 2015. 12. 037.LU Heng,FU Xiao,HE Yinan,et al. Cultivated land information extraction from high resolution UAV images based on transfer learning[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2015,46(12):274-279,284.(in Chinese)
[6]易超人,邓燕妮.多通道卷积神经网络图像识别方法[J].河南科技大学学报(自然科学版),2017,38(3):41-44.YI Chaoren,DENG Yanni. Multi-channel convolutional neural network image recognition method[J]. Journal of Henan University of Science&Technology(Natural Science),2017,38(3):41-44.(in Chinese)
[7]李云飞,符冉迪,金炜,等.多通道卷积的图像超分辨率方法[J].中国图象图形学报,2017,22(12):1690-1700.LI Yunfei,FU Randi,JIN Wei,et al. Image super-resolution using multi-channel convolution[J]. Journal of Image and Graphics,2017,22(12):1690-1700.(in Chinese)
[8] HE Kaiming,ZHANG Xiangyu,REN Shaoqing,et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(9):1904-1916.
[9] DONAHUE J,JIA Y,VINYALS O,et al. De CAF:a deep convolutional activation feature for generic visual recognition[C]∥International Conference on International Conference on Machine Learning,2014,32:647-665.
[10]韩星烁,林伟.深度卷积神经网络在图像识别算法中的研究与实现[J].微型机与应用,2017,36(21):54-56.HAN Xingshuo,LIN Wei. Design and implementation of image recognition based on convolution neural network[J].Information and Technology and Network Security,2017,36(21):54-56.(in Chinese)
[11]孙钰,韩京冶,陈志泊,等.基于深度学习的大棚及地膜农田无人机航拍监测方法[J/OL].农业机械学报,2018,49(2):133-140. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? file_no=20180218&flag=1. DOI:10. 6041/j.issn. 1000-1298. 2018. 02. 018.SUN Yu,HAN Jingye,CHEN Zhibo,et al. Monitoring method for UAV image of greenhouse and plastic-mulched landcover based on deep learning[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2018,49(2):133-140.(in Chinese)
[12]刘明月.克里雅河下游水系特征及对植被的影响[D].乌鲁木齐:新疆大学,2013.LIU Mingyue. Water characteristics of Keriya river downstream and impact on vagetaion[D]. Urumqi:Xinjiang University,2013.(in Chinese)
[13]阿斯姆古丽·阿纳耶提.沙漠腹地绿洲达理雅博依原住民居住环境空间特征及环境适宜性分析[D].乌鲁木齐:新疆大学,2014.ASIMUGULI Anayeti. Spatial characteristics of aboriginal living conditional and environmental suitability analysis of Daryaboyi oasis[D]. Urumqi:Xinjiang University,2014.(in Chinese)
[14]段萌,王功鹏,牛常勇.基于卷积神经网络的小样本图像识别方法[J].计算机工程与设计,2018,39(1):224-229.DUAN Meng,WANG Gongpeng,NIU Changyong. Method of small sample size image recognition based on convolution neural network[J]. Computer Engineering and Design,2018,39(1):224-229.(in Chinese)
[15] ZEILER M D,FERGUS R. Visualizing and understanding convolutional networks[C]∥European Conference on Computer Vision,2014:818-833.
[16]陈小娥,杨薇薇.基于深度学习的车标识别算法的研究与实现[J].长春工程学院学报(自然科学版),2017,18(2):117-120.CHEN Xiaoe,YANG Weiwei. The study and implement of vehicle logo recognition algorithm based on deep learning[J].Journal of Changchun Institute of Technology(Natural Science Edition),2017,18(2):117-120.(in Chinese)
[17]鲁恒,付萧,刘超,等.基于低空遥感与迁移学习的土地利用信息快速制图方法[J/OL].农业机械学报,2016,47(11):262-269. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? file_no=20161136&flag=1. DOI:10.6041/j. issn. 1000-1298. 2016. 11. 036.LU Heng,FU Xiao,LIU Chao,et al. Landuse information quick mapping based on low altitude remote sensing technology and transfer learning[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2016,47(11):262-269.(in Chinese)
[18] ZHOU Xun,SUN Zhongping,LIU Suhong,et al. A method for extracting the leaf litter distribution area in forest using chip feature[J]. International Journal of Remote Sensing,2018,39(15-16):5310-5329.
[19]孙超,吕俊伟,李健伟,等.基于去卷积的快速图像超分辨率方法[J].光学学报,2017,37(12):142-152.SUN Chao,LJunwei,LI Jianwei,et al. Method of rapid image super-resolution based on deconvolution[J]. Acta Optica Sinica,2017,37(12):142-152.(in Chinese)
[20]李现国,孙叶美,杨彦利,等.基于中间层监督卷积神经网络的图像超分辨率重建[J].中国图象图形学报,2018,23(7):984-993.LI Xianguo,SUN Yemei,YANG Yanli,et al. Image super-resolution reconstruction based on intermediate supervision convolutional neural networks[J]. Journal of Image and Graphics,2018,23(7):984-993.(in Chinese)
[2]张同瑞,赵庚星,高明秀,等.基于近地多光谱和OLI影像的黄河三角洲冬小麦种植区盐分估算及遥感反演——以山东省垦利县和无棣县为例[J].自然资源学报,2016,31(6):1051-1060.ZHANG Tongrui,ZHAO Gengxing,GAO Mingxiu,et al. Soil salinity estimation and remote sensing inversion based on nearground multispectral and TM imagery in winter wheat growing area in the Yellow River Delta—case study in Kenli County and Wudi County,Shandong Province[J]. Journal of Natural Resources,2016,31(6):1051-1060.(in Chinese)
[3]肖海燕,曾辉,昝启杰,等.基于高光谱数据和专家决策法提取红树林群落类型信息[J].遥感学报,2007,11(4):531-537.XIAO Haiyan,ZENG Hui,ZAN Qijie,et al. Decision tree model in extraction of mangrove community information using hyperspectral image data[J]. Journal of Remote Sensing,2007,11(4):531-537.(in Chinese)
[4]杜欣,黄晓霞,李红旮,等.基于投影寻踪学习网络算法的植物群落高分遥感分类研究[J].地球信息科学学报,2016,18(1):124-132.DU Xin,HUANG Xiaoxia,LI Hongga,et al. Research on classification of plant community using projection pursuit learning network algorithm on high resolution remote sensing images[J]. Journal of Geo-information Science,2016,18(1):124-132.(in Chinese)
[5]鲁恒,付萧,贺一楠,等.基于迁移学习的无人机影像耕地信息提取方法[J/OL].农业机械学报,2015,46(12):274-279,284. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? file_no=20151237&flag=1. DOI:10. 6041/j. issn.1000-1298. 2015. 12. 037.LU Heng,FU Xiao,HE Yinan,et al. Cultivated land information extraction from high resolution UAV images based on transfer learning[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2015,46(12):274-279,284.(in Chinese)
[6]易超人,邓燕妮.多通道卷积神经网络图像识别方法[J].河南科技大学学报(自然科学版),2017,38(3):41-44.YI Chaoren,DENG Yanni. Multi-channel convolutional neural network image recognition method[J]. Journal of Henan University of Science&Technology(Natural Science),2017,38(3):41-44.(in Chinese)
[7]李云飞,符冉迪,金炜,等.多通道卷积的图像超分辨率方法[J].中国图象图形学报,2017,22(12):1690-1700.LI Yunfei,FU Randi,JIN Wei,et al. Image super-resolution using multi-channel convolution[J]. Journal of Image and Graphics,2017,22(12):1690-1700.(in Chinese)
[8] HE Kaiming,ZHANG Xiangyu,REN Shaoqing,et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(9):1904-1916.
[9] DONAHUE J,JIA Y,VINYALS O,et al. De CAF:a deep convolutional activation feature for generic visual recognition[C]∥International Conference on International Conference on Machine Learning,2014,32:647-665.
[10]韩星烁,林伟.深度卷积神经网络在图像识别算法中的研究与实现[J].微型机与应用,2017,36(21):54-56.HAN Xingshuo,LIN Wei. Design and implementation of image recognition based on convolution neural network[J].Information and Technology and Network Security,2017,36(21):54-56.(in Chinese)
[11]孙钰,韩京冶,陈志泊,等.基于深度学习的大棚及地膜农田无人机航拍监测方法[J/OL].农业机械学报,2018,49(2):133-140. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? file_no=20180218&flag=1. DOI:10. 6041/j.issn. 1000-1298. 2018. 02. 018.SUN Yu,HAN Jingye,CHEN Zhibo,et al. Monitoring method for UAV image of greenhouse and plastic-mulched landcover based on deep learning[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2018,49(2):133-140.(in Chinese)
[12]刘明月.克里雅河下游水系特征及对植被的影响[D].乌鲁木齐:新疆大学,2013.LIU Mingyue. Water characteristics of Keriya river downstream and impact on vagetaion[D]. Urumqi:Xinjiang University,2013.(in Chinese)
[13]阿斯姆古丽·阿纳耶提.沙漠腹地绿洲达理雅博依原住民居住环境空间特征及环境适宜性分析[D].乌鲁木齐:新疆大学,2014.ASIMUGULI Anayeti. Spatial characteristics of aboriginal living conditional and environmental suitability analysis of Daryaboyi oasis[D]. Urumqi:Xinjiang University,2014.(in Chinese)
[14]段萌,王功鹏,牛常勇.基于卷积神经网络的小样本图像识别方法[J].计算机工程与设计,2018,39(1):224-229.DUAN Meng,WANG Gongpeng,NIU Changyong. Method of small sample size image recognition based on convolution neural network[J]. Computer Engineering and Design,2018,39(1):224-229.(in Chinese)
[15] ZEILER M D,FERGUS R. Visualizing and understanding convolutional networks[C]∥European Conference on Computer Vision,2014:818-833.
[16]陈小娥,杨薇薇.基于深度学习的车标识别算法的研究与实现[J].长春工程学院学报(自然科学版),2017,18(2):117-120.CHEN Xiaoe,YANG Weiwei. The study and implement of vehicle logo recognition algorithm based on deep learning[J].Journal of Changchun Institute of Technology(Natural Science Edition),2017,18(2):117-120.(in Chinese)
[17]鲁恒,付萧,刘超,等.基于低空遥感与迁移学习的土地利用信息快速制图方法[J/OL].农业机械学报,2016,47(11):262-269. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? file_no=20161136&flag=1. DOI:10.6041/j. issn. 1000-1298. 2016. 11. 036.LU Heng,FU Xiao,LIU Chao,et al. Landuse information quick mapping based on low altitude remote sensing technology and transfer learning[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2016,47(11):262-269.(in Chinese)
[18] ZHOU Xun,SUN Zhongping,LIU Suhong,et al. A method for extracting the leaf litter distribution area in forest using chip feature[J]. International Journal of Remote Sensing,2018,39(15-16):5310-5329.
[19]孙超,吕俊伟,李健伟,等.基于去卷积的快速图像超分辨率方法[J].光学学报,2017,37(12):142-152.SUN Chao,LJunwei,LI Jianwei,et al. Method of rapid image super-resolution based on deconvolution[J]. Acta Optica Sinica,2017,37(12):142-152.(in Chinese)
[20]李现国,孙叶美,杨彦利,等.基于中间层监督卷积神经网络的图像超分辨率重建[J].中国图象图形学报,2018,23(7):984-993.LI Xianguo,SUN Yemei,YANG Yanli,et al. Image super-resolution reconstruction based on intermediate supervision convolutional neural networks[J]. Journal of Image and Graphics,2018,23(7):984-993.(in Chinese)