基于Adaboost的高分遥感影像自动变化检测方法
|
摘要
基于监督分类的高分辨率遥感影像变化检测需要大量人工标注,且单个监督分类器难以适应高分影像中复杂多样的地表变化信息提取,检测结果中"椒盐噪声"严重、变化图斑破碎。因此,本文提出一种基于Adaboost集成算法、自动标注训练样本的变化检测方法。首先利用非监督分类方法完成变化初检,接着在初检结果中进行"非等距"区间采样自动获取均匀分布的训练样本;然后以Adaboost算法为集成框架,选择决策树桩、Logistic回归和k NN作为弱分类器,构建一种混合分类器集成系统,充分挖掘和利用高分影像中的空间信息以提升分类精度和分类器泛化能力,最后利用SLIC分割算法和空间邻域信息对像元级检测结果进行空间约束滤波,进一步提升变化检测精度。为验证本文方法的有效性,选取SPOT-5和WorldView-2影像为实验数据,结果表明本文方法能有效降低训练样本人工标注成本、提高变化检测精度。
Human annotation is a massive labor cost for the training sample selection process when applying any kind of supervised learning algorithm for change detection based on high-resolution remotely sensed satellite images. It is limited and unreasonable to use just one single sort of classifier generated from a supervised algorithm to extract change information of variety from the time-series images both in completeness and accuracy, let alone the inevitable salt-and-pepper noise and tiny patches falsely detected which turn out to be ubiquitous in and out of geographical entities. To tackle with problems mentioned above, a change detection approach based on a new automatic training sample annotation strategy and an improved Adaboost ensemble learning algorithm was proposed. At first, the unsupervised change detection algorithm CVA was applied to generate a low-level change detection result as referencing labels for further annotation, then the low-level result was divided into several parts with different intervals to ensure the automatic acquisition of the evenly distributed training samples with confidence. Furthermore, decision stump, logistic regression and kNN were employed as the weak classifiers to construct a hybrid multi-classifiers ensemble system with the help of the improved Adaboost algorithm, which would effectively promote the classification accuracy and generalization capacity of weak classifiers by sufficiently mining and making use of the spatial information with potential values. Finally, the SLIC segmentation algorithm was implemented in the difference image, and the segmentation border information was combined with spatial contextual information to build up a dual-filter for spatial constraint aiming at decreasing the omission rate and the false alarm rate of the detection results. To verify the validity of the proposed method, we conducted experiments using two datasets of multispectral images collected by SPOT-5 and WorldView-2. Experimental results indicated that the proposed method would significantly lower the labor costs of training sample annotation and demonstrated superiority compared with four other methods in accuracy.
Human annotation is a massive labor cost for the training sample selection process when applying any kind of supervised learning algorithm for change detection based on high-resolution remotely sensed satellite images. It is limited and unreasonable to use just one single sort of classifier generated from a supervised algorithm to extract change information of variety from the time-series images both in completeness and accuracy, let alone the inevitable salt-and-pepper noise and tiny patches falsely detected which turn out to be ubiquitous in and out of geographical entities. To tackle with problems mentioned above, a change detection approach based on a new automatic training sample annotation strategy and an improved Adaboost ensemble learning algorithm was proposed. At first, the unsupervised change detection algorithm CVA was applied to generate a low-level change detection result as referencing labels for further annotation, then the low-level result was divided into several parts with different intervals to ensure the automatic acquisition of the evenly distributed training samples with confidence. Furthermore, decision stump, logistic regression and kNN were employed as the weak classifiers to construct a hybrid multi-classifiers ensemble system with the help of the improved Adaboost algorithm, which would effectively promote the classification accuracy and generalization capacity of weak classifiers by sufficiently mining and making use of the spatial information with potential values. Finally, the SLIC segmentation algorithm was implemented in the difference image, and the segmentation border information was combined with spatial contextual information to build up a dual-filter for spatial constraint aiming at decreasing the omission rate and the false alarm rate of the detection results. To verify the validity of the proposed method, we conducted experiments using two datasets of multispectral images collected by SPOT-5 and WorldView-2. Experimental results indicated that the proposed method would significantly lower the labor costs of training sample annotation and demonstrated superiority compared with four other methods in accuracy.
引文
[1] Singh A. Digital change detection techniques using remotely-sensed data[J]. International Journal of Remote Sensing, 1989,10(6):989-1003.
[2] Gu Y J, Jin Z, Chiu S C. Active learning combining uncertainty and diversity for multi-class image classification[J]. IET Computer Vision, 2015,9(3):400-407
[3]吴田军,骆剑承,夏列钢,等.迁移学习支持下的遥感影像对象级分类样本自动选择方法[J].测绘学报,2014,43(9):908-916.[Wu T J, Luo J C, Xia L G, et al. An automatic sample collection method for object-oriented classification of remotely sensed imageries baed on transfer learning[J]. Acta Geodaetica et Cartographica Sinica, 2014,43(9):908-916.]
[4] Huo C, Zhou Z, Lu H., et al. Fast object-level change detection for VHR images[J]. IEEE Geoscience and Remote Sensing Letters, 2010,7(1):118-122.
[5]张鑫龙,陈秀万,李飞,等.高分辨率遥感影像的深度学习变化检测方法[J].测绘学报,2017,46(8):999-1008.[Zhang X L, Chen J W, Li F, Et al. Change detection method for high resolution remote sensing images using deep learning[J]. Acta Geodaetica et Cartographica Sinica,2017,46(8):999-1008.]
[6] Tan K, Jin X, Antonio P, et al. Automatic change detection in high-resolution remote sensing images by using a multiple classifier system and spectral-spatial features[J].IEEE Journal of Selected Topics in Applied Earth Observations And Remote Sensing, 2016,9(8):3439-3451.
[7] Cao G, Wang B, H. Xavier, et al. A new difference image creation method based on deep neural networks for change detection in remote-sensing images[J]. International Journal of Remote Sensing, 2017,38(23):7161-7175.
[8]魏立飞,钟燕飞,张良培,等.遥感影像融合的自适应变化检测[J].遥感学报,2010,14(6):1196-1211.[Wei L F,Zhong Y F, Zhang L P, et al. Adaptive change method of remote sensing image fusion[J]. Journal of Remote Sensing, 2010,14(6):1196-1211.]
[9] Freund Y,Schapire R. A decision-theoretic generalization of on-line learning and an application to boosting[J].Journal of Computer and System Sciences, 1997,55(1):119-139.
[10]龚健雅,姚璜,沈欣.利用AdaBoost算法进行高分辨率影像的面向对象分类[J].武汉大学学报·信息科学版,2010,35(12):1440-1443.[Gong J Y, Yao H, Shen X. Object-oriented classification of high spatial-resolution remote sensing imagery based on Ada Boost[J]. Geomatics and Information Science of Wuhan University, 2010,35(12):1440-1443.]
[11] Woo D M,Do V D. Post-Classification change detection of high resolution satellite images using AdaBoost classifier[J]. Information Technology&Computer Science,2015,117:34-38.
[12] Chen Y, Dou P, Yang X. Improving land use/cover classification with a multiple classifier system using AdaBoost integration technique[J]. Remote Sensing, 2017,9(10):1055-1074.
[13] Schapire R E,Singer Y. Improved boosting algorithms using confidence-rated predictions[J]. Machine Learning,1999,37(3):297-336.
[14] Wu S Q, H. Nagahashi. Analysis of generalization ability for different AdaBoost variants based on classification and regression trees[J]. Journal of Electrical and Computer Engineering, 2015,2015:1-17.
[15] Zhang C, Ma Y. Ensemble machine learning:methods and applications[M]. New York:Springer-Verlag, 2012,88.
[16] Mancini A, Frontoni E, Zingaretti P. Road change detection from multi-spectral aerial data[C]//International Conference on Pattern Recognition, 2010:448-451.
[17] Le H L, Woo D M. Combination of textural features for the improvement of terrain classification and change detection[J]. International Journal of Software Engineering And Its Applications, 2015,9(5):145-154.
[18] Peter H,李锐,李鹏,等.机器学习实战[M].北京:人民邮电出版社,2013.[Peter H., Li R, Li P, et al. Machine learning in action[M]. Beijing:Posts And Telecom Press, 2013.]
[19] Barreto T L M, Rosa R A S, Wimmer C, et al. Classification of detected changes from multitemporal high-res Xband SAR images:intensity and texture descriptors from superpixels[J]. IEEE Journal of Selected Topics In Applied Earth Observations and Remote Sensing, 2016,9(12):5436-5448.
[20] Toro C, Martín C G, PedreroáG, et al. Superpixel-based roughness measure for multispectral satellite image segmentation[J]. Remote Sensing, 2015,7(11):14620-14645.
[21] Zhou L, Cao G, Li Y, et al. Change detection based on conditional random field with region connection constraints in high-resolution remote sensing images[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016,9(8):3478-3488.
[2] Gu Y J, Jin Z, Chiu S C. Active learning combining uncertainty and diversity for multi-class image classification[J]. IET Computer Vision, 2015,9(3):400-407
[3]吴田军,骆剑承,夏列钢,等.迁移学习支持下的遥感影像对象级分类样本自动选择方法[J].测绘学报,2014,43(9):908-916.[Wu T J, Luo J C, Xia L G, et al. An automatic sample collection method for object-oriented classification of remotely sensed imageries baed on transfer learning[J]. Acta Geodaetica et Cartographica Sinica, 2014,43(9):908-916.]
[4] Huo C, Zhou Z, Lu H., et al. Fast object-level change detection for VHR images[J]. IEEE Geoscience and Remote Sensing Letters, 2010,7(1):118-122.
[5]张鑫龙,陈秀万,李飞,等.高分辨率遥感影像的深度学习变化检测方法[J].测绘学报,2017,46(8):999-1008.[Zhang X L, Chen J W, Li F, Et al. Change detection method for high resolution remote sensing images using deep learning[J]. Acta Geodaetica et Cartographica Sinica,2017,46(8):999-1008.]
[6] Tan K, Jin X, Antonio P, et al. Automatic change detection in high-resolution remote sensing images by using a multiple classifier system and spectral-spatial features[J].IEEE Journal of Selected Topics in Applied Earth Observations And Remote Sensing, 2016,9(8):3439-3451.
[7] Cao G, Wang B, H. Xavier, et al. A new difference image creation method based on deep neural networks for change detection in remote-sensing images[J]. International Journal of Remote Sensing, 2017,38(23):7161-7175.
[8]魏立飞,钟燕飞,张良培,等.遥感影像融合的自适应变化检测[J].遥感学报,2010,14(6):1196-1211.[Wei L F,Zhong Y F, Zhang L P, et al. Adaptive change method of remote sensing image fusion[J]. Journal of Remote Sensing, 2010,14(6):1196-1211.]
[9] Freund Y,Schapire R. A decision-theoretic generalization of on-line learning and an application to boosting[J].Journal of Computer and System Sciences, 1997,55(1):119-139.
[10]龚健雅,姚璜,沈欣.利用AdaBoost算法进行高分辨率影像的面向对象分类[J].武汉大学学报·信息科学版,2010,35(12):1440-1443.[Gong J Y, Yao H, Shen X. Object-oriented classification of high spatial-resolution remote sensing imagery based on Ada Boost[J]. Geomatics and Information Science of Wuhan University, 2010,35(12):1440-1443.]
[11] Woo D M,Do V D. Post-Classification change detection of high resolution satellite images using AdaBoost classifier[J]. Information Technology&Computer Science,2015,117:34-38.
[12] Chen Y, Dou P, Yang X. Improving land use/cover classification with a multiple classifier system using AdaBoost integration technique[J]. Remote Sensing, 2017,9(10):1055-1074.
[13] Schapire R E,Singer Y. Improved boosting algorithms using confidence-rated predictions[J]. Machine Learning,1999,37(3):297-336.
[14] Wu S Q, H. Nagahashi. Analysis of generalization ability for different AdaBoost variants based on classification and regression trees[J]. Journal of Electrical and Computer Engineering, 2015,2015:1-17.
[15] Zhang C, Ma Y. Ensemble machine learning:methods and applications[M]. New York:Springer-Verlag, 2012,88.
[16] Mancini A, Frontoni E, Zingaretti P. Road change detection from multi-spectral aerial data[C]//International Conference on Pattern Recognition, 2010:448-451.
[17] Le H L, Woo D M. Combination of textural features for the improvement of terrain classification and change detection[J]. International Journal of Software Engineering And Its Applications, 2015,9(5):145-154.
[18] Peter H,李锐,李鹏,等.机器学习实战[M].北京:人民邮电出版社,2013.[Peter H., Li R, Li P, et al. Machine learning in action[M]. Beijing:Posts And Telecom Press, 2013.]
[19] Barreto T L M, Rosa R A S, Wimmer C, et al. Classification of detected changes from multitemporal high-res Xband SAR images:intensity and texture descriptors from superpixels[J]. IEEE Journal of Selected Topics In Applied Earth Observations and Remote Sensing, 2016,9(12):5436-5448.
[20] Toro C, Martín C G, PedreroáG, et al. Superpixel-based roughness measure for multispectral satellite image segmentation[J]. Remote Sensing, 2015,7(11):14620-14645.
[21] Zhou L, Cao G, Li Y, et al. Change detection based on conditional random field with region connection constraints in high-resolution remote sensing images[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016,9(8):3478-3488.