利用多属性剖面概率融合的高光谱影像分类
|
摘要
针对单一的属性剖面特征难以全面反映地物特性的问题,提出一种融合多种属性剖面特征的高光谱影像分类方法。首先提取高光谱影像4种形态学属性剖面特征,接着基于4组剖面特征对高光谱影像进行分类,获得样本的预测类别和后验概率估计值;在此基础上,计算不同特征的分类可靠性以及重要度权值,二者结合建立基于概率的决策融合模型,获得高光谱影像的最终分类结果。高光谱影像分类实验表明,所提融合算法的性能不但优于使用单个属性剖面特征的情况,也优于其他多种融合算法。
A new method for hyperspectral image classification is proposed.In the method,multi-attribute profiles are used and fused by aprobabilistic fusion model.First,four types of attribute profile features are extracted from the reduced images which are generated by dimensionality reduction.Next,based on the four types of features,four classification results and the corresponding estimation of posterior probability are obtained.Then,the reliability and important weight of each feature are calculated based on the predicted labels and posterior probabilities.Finally,aprobabilistic fusion model is constructed using the reliabilities and weights of features,and the final classification results are obtained.The experiments of hyperspectral image classification show that the proposed method can not only achieve better performance than the methods using any of single features,but also outperform other fusion methods.
A new method for hyperspectral image classification is proposed.In the method,multi-attribute profiles are used and fused by aprobabilistic fusion model.First,four types of attribute profile features are extracted from the reduced images which are generated by dimensionality reduction.Next,based on the four types of features,four classification results and the corresponding estimation of posterior probability are obtained.Then,the reliability and important weight of each feature are calculated based on the predicted labels and posterior probabilities.Finally,aprobabilistic fusion model is constructed using the reliabilities and weights of features,and the final classification results are obtained.The experiments of hyperspectral image classification show that the proposed method can not only achieve better performance than the methods using any of single features,but also outperform other fusion methods.
引文
[1]MELGANI F,BRUZZONE L.Classification of hyperspectral remote sensing images with support vector machines[J].IEEE Transactions on Geoscience &Remote Sensing,2004,42(8):1778-1790.
[2]赵萍,傅云飞,郑刘根,等.基于分类回归树分析的遥感影像土地利用/覆被分类研究[J].遥感学报,2005,9(6):708-716.
[3]HAM J,CHEN Y,CRAWFORD M M,et al.Investigation of the random forest framework for classification of hyperspectral data[J].IEEE Transactions on Geoscience &Remote Sensing,2005,43(3):492-501.
[4]陈玉敏.基于神经网络的遥感影像分类研究[J].测绘地理信息,2002,27(3):6-8.
[5]田彦平,陶超,邹峥嵘,等.主动学习与图的半监督相结合的高光谱影像分类[J].测绘学报,2015,44(8):919-926.
[6]王立国,孙杰,肖倩.结合空-谱信息的高光谱图像分类方法[J].黑龙江大学自然科学学报,2010,27(6):788-791.
[7]LI J,MARPU P R,PLAZA A,et al.Generalized composite kernel framework for hyperspectral image classification[J].IEEE Transactions on Geoscience &Remote Sensing,2013,51(9):4816-4829.
[8]BENEDIKTSSON J A.Extended profiles with morphological attribute filters for the analysis of hyperspectral data[J].International Journal of Remote Sensing,2010,31(22):5975-5991.
[9]王扣准,黄睿.利用多属性剖面与双边滤波的高光谱影像分类[J].遥感信息,2016,31(6):104-109.
[10]MURA M D,BENEDIKTSSON J A,WASKE B,et al.Morphological attribute profiles for the analysis of very high resolution images[J].IEEE Transactions on Geoscience &Remote Sensing,2010,48(10):3747-3762.
[11]ZHONG Y,ZHU Q,ZHANG L.Scene classification based on the multifeature fusion probabilistic topic model for high spatial resolution remote sensing imagery[J].IEEE Transactions on Geoscience &Remote Sensing,2015,53(11):6207-6222.
[12]ZHANG L,ZHANG L,TAO D,et al.On combining multiple features for hyperspectral remote sensing image classification[J].IEEE Transactions on Geoscience &Remote Sensing,2012,50(3):879-893.
[13]HUANG X,ZHANG L.An SVM ensemble approach combining spectral,structural,and semantic features for the classification of high-resolution remotely sensed imagery[J].IEEE Transactions on Geoscience &Remote Sensing,2013,51(1):257-272.
[14]张春森,郑艺惟,黄小兵,等.高光谱影像光谱-空间多特征加权概率融合分类[J].测绘学报,2015,44(8):909-918.
[2]赵萍,傅云飞,郑刘根,等.基于分类回归树分析的遥感影像土地利用/覆被分类研究[J].遥感学报,2005,9(6):708-716.
[3]HAM J,CHEN Y,CRAWFORD M M,et al.Investigation of the random forest framework for classification of hyperspectral data[J].IEEE Transactions on Geoscience &Remote Sensing,2005,43(3):492-501.
[4]陈玉敏.基于神经网络的遥感影像分类研究[J].测绘地理信息,2002,27(3):6-8.
[5]田彦平,陶超,邹峥嵘,等.主动学习与图的半监督相结合的高光谱影像分类[J].测绘学报,2015,44(8):919-926.
[6]王立国,孙杰,肖倩.结合空-谱信息的高光谱图像分类方法[J].黑龙江大学自然科学学报,2010,27(6):788-791.
[7]LI J,MARPU P R,PLAZA A,et al.Generalized composite kernel framework for hyperspectral image classification[J].IEEE Transactions on Geoscience &Remote Sensing,2013,51(9):4816-4829.
[8]BENEDIKTSSON J A.Extended profiles with morphological attribute filters for the analysis of hyperspectral data[J].International Journal of Remote Sensing,2010,31(22):5975-5991.
[9]王扣准,黄睿.利用多属性剖面与双边滤波的高光谱影像分类[J].遥感信息,2016,31(6):104-109.
[10]MURA M D,BENEDIKTSSON J A,WASKE B,et al.Morphological attribute profiles for the analysis of very high resolution images[J].IEEE Transactions on Geoscience &Remote Sensing,2010,48(10):3747-3762.
[11]ZHONG Y,ZHU Q,ZHANG L.Scene classification based on the multifeature fusion probabilistic topic model for high spatial resolution remote sensing imagery[J].IEEE Transactions on Geoscience &Remote Sensing,2015,53(11):6207-6222.
[12]ZHANG L,ZHANG L,TAO D,et al.On combining multiple features for hyperspectral remote sensing image classification[J].IEEE Transactions on Geoscience &Remote Sensing,2012,50(3):879-893.
[13]HUANG X,ZHANG L.An SVM ensemble approach combining spectral,structural,and semantic features for the classification of high-resolution remotely sensed imagery[J].IEEE Transactions on Geoscience &Remote Sensing,2013,51(1):257-272.
[14]张春森,郑艺惟,黄小兵,等.高光谱影像光谱-空间多特征加权概率融合分类[J].测绘学报,2015,44(8):909-918.