基于CAR-Siamese网络的高分辨率遥感图像建筑物变化检测

doi:10.7523/j.ucas.2021.0035

摘要/Abstract

摘要： 准确提取建筑物变化区域对城乡规划、地理国情监测、城市扩张分析有着重要意义。传统遥感变化检测方法难以适应遥感图像复杂场景下的变化检测任务的要求。近年来广泛应用于计算机视觉领域的深度学习变化检测算法相对于传统方法在效率和精度上有明显提升。然而遥感图像上建筑物特征丰富、变化多样，且建筑物变化样本获取难度大，导致现有深度学习模型在建筑物变化检测任务上精度受限。针对这一问题，提出变化注意力残差孪生网络(CAR-siamese net)，增强不同尺度下图像信息的共享交流，充分学习建筑物的变化特征，同时，提出建筑物语义分割样本预训练策略，有效利用现有建筑物分割样本，最终提升了变化检测网络对建筑物变化的解译能力。以北京昌平区影像为底图制作建筑物变化检测数据集，在该数据集和Levir-CD公开数据集上的实验结果表明，该方法能有效提高建筑物变化检测精度。

关键词: 变化检测, 建筑物, 深度学习, 卷积神经网络, 孪生网络, 变化注意力残差

Abstract: Accurately extracting building change regions is of great significance to urban and rural planning, geographic national conditions monitoring, and urban expansion analysis. Traditional remote sensing change detection methods are difficult to adapt to the change detection tasks in complex scenes of remote sensing images. In recent years, deep learning change detection algorithms, which have been widely used in the field of computer vision, have significantly improved efficiency and accuracy compared to traditional methods. However, the features of buildings on remote sensing images are rich and varied, and it is difficult to obtain samples of building changes, which leads to the limited accuracy of existing deep learning models in building change detection tasks. This paper proposes a change attention residual siamese network (CAR-siamese net), which enhances the interaction of image information at different scales, and fully learns the change features of buildings. In addition, a pre-training strategy is proposed in this paper to effectively use building segmentation samples, and the ability of the change detection network to interpret building changes is improved. In this paper, a building change detection data set is made based on images of Changping District, Beijing. Experimental results on this data set and Levir-CD public data set show that the method in this paper can effectively improve the accuracy of building change detection.

Key words: change detection, building, deep learning, convolutional neural networks, siamese networks, change attention residual

中图分类号:

TP751

姚沐风, 昝露洋, 李柏鹏, 李庆亭, 陈正超. 基于CAR-Siamese网络的高分辨率遥感图像建筑物变化检测[J]. 中国科学院大学学报, 2023, 40(3): 380-387.

YAO Mufeng, ZAN Luyang, LI Baipeng, LI Qingting, CHEN Zhengchao. Building change detection from remote sensing images using CAR-Siamese net[J]. Journal of University of Chinese Academy of Sciences, 2023, 40(3): 380-387.

参考文献

[1] 张良培, 武辰. 多时相遥感影像变化检测的现状与展望[J]. 测绘学报, 2017, 46(10):1447-1459.
[2] 马建文, 田国良, 王长耀, 等. 遥感变化检测技术发展综述[J]. 地球科学进展, 2004, 19(2):192-196.
[3] 杜培军, 王欣, 蒙亚平, 等. 面向地理国情监测的变化检测与地表覆盖信息更新方法[J]. 地球信息科学学报, 2020, 22(4):857-866.
[4] Lu L L, Guo H D, Corbane C, et al. Urban sprawl in provincial capital cities in China:evidence from multi-temporal urban land products using Landsat data[J]. Science Bulletin, 2019, 64(14):955-957.
[5] Huang X, Zhang L P, Zhu T T. Building change detection from multitemporal high-resolution remotely sensed images based on a morphological building index[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(1):105-115.
[6] Tang Y Q, Huang X, Zhang L P. Fault-tolerant building change detection from urban high-resolution remote sensing imagery[J]. IEEE Geoscience and Remote Sensing Letters, 2013, 10(5):1060-1064.
[7] Huang X, Zhu T T, Zhang L P, et al. A novel building change index for automatic building change detection from high-resolution remote sensing imagery[J]. Remote Sensing Letters, 2014, 5(8):713-722.
[8] Sofina N, Ehlers M. Building change detection using high resolution remotely sensed data and GIS[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(8):3430-3438.
[9] Gueguen L, Pesaresi M, Ehrlich D, et al. Urbanization detection by a region based mixed information change analysis between built-up indicators[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2013, 6(6):2410-2420.
[10] Huang X, Wen D W, Li J Y, et al. Multi-level monitoring of subtle urban changes for the megacities of China using high-resolution multi-view satellite imagery[J]. Remote Sensing of Environment, 2017, 196:56-75.
[11] Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6):84-90.
[12] Shelhamer E, Long J, Darrell T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4):640-651.
[13] Ren S Q, He K M, Girshick R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.
[14] Gong J Q, Hu X Y, Pang S Y, et al. Patch matching and dense CRF-based co-refinement for building change detection from bi-temporal aerial images[J]. Sensors, 2019, 19(7):1557.
[15] Ji S P, Shen Y Y, Lu M, et al. Building instance change detection from large-scale aerial images using convolutional neural networks and simulated samples[J]. Remote Sensing, 2019, 11(11):1343.
[16] Caye Daudt R, Le Saux B, Boulch A. Fully convolutional siamese networks for change detection[C]//International Conference on Image Processing. Athens, Greece:IEEE, 2018:4063-4067.
[17] Zhan Y, Fu K, Yan M L, et al. Change detection based on deep Siamese convolutional network for optical aerial images[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(10):1845-1849.
[18] El Amin A M, Liu Q J, Wang Y H. Zoom out CNNs features for optical remote sensing change detection[C]//International Conference on Image, Vision and Computing. Chengdu, China:IEEE, 2017:812-817.
[19] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[C]//International Conference on Learning Representations. San Diego, CA, USA:Computational and Biological Learning Society, 2015:1-14.
[20] Caye Daudt R, Le Saux B, Boulch A, et al. Multitask learning for large-scale semantic change detection[J]. Computer Vision and Image Understanding, 2019, 187:102783.
[21] Bao T F, Fu C Q, Fang T, et al. PPCNET:a combined patch-level and pixel-level end-to-end deep network for high-resolution remote sensing image change detection[J]. IEEE Geoscience and Remote Sensing Letters, 2020, 17(10):1797-1801.
[22] Jiang H W, Hu X Y, Li K, et al. PGA-SiamNet:pyramid feature-based attention-guided siamese network for remote sensing orthoimagery building change detection[J]. Remote Sensing, 2020, 12(3):484.
[23] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, NV, USA:IEEE, 2016:770-778.
[24] Chen H, Shi Z W. A spatial-temporal attention-based method and a new dataset for remote sensing image change detection[J]. Remote Sensing, 2020, 12(10):1662.
[25] Peng X L, Zhong R F, Li Z, et al. Optical remote sensing image change detection based on attention mechanism and image difference[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, PP(99):1-12.
[26] Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Salt Lake City, UT, USA:IEEE, 2018:7132-7141.