基于深度学习模型的遥感图像分割方法

doi:10.11772/j.issn.1001-9081.2019030529

摘要/Abstract

摘要： 利用遥感图像快速准确地检测地物信息是当前的研究热点。针对遥感图像地表物的传统人工目视解译分割方法效率低下和现有基于深度学习的遥感图像分割算法在复杂场景下准确率不高、背景噪声多的问题，提出一种基于改进的U-net架构与全连接条件随机场的图像分割算法。首先，融合VGG16和U-net构建新的网络模型，以有效提取具有高背景复杂度的遥感图像特征；然后，通过选取适当的激活函数和卷积方式，在提高图像分割准确率的同时显著降低模型预测时间；最后，在保证分割精度的基础上，使用全连接条件随机场进一步优化分割结果，以获得更加细致的分割边缘。在ISPRS提供的标准数据集Potsdam上进行的仿真测试表明，相较于U-net，所提算法的准确率、召回率和均交并比（MIoU）分别提升了15.06个百分点、29.11个百分点和0.3662，平均绝对误差（MAE）降低了0.02892。实验结果验证了该算法具备有效性和鲁棒性，是一种有效的遥感图像地表物提取算法。

关键词: 深度学习, 卷积神经网络, 深度可分离卷积, 全连接条件随机场

Abstract: To detect surface object information quickly and accurately by using remote sensing images is a current research hot spot. In order to solve the problems of inefficiency of the traditional manual visual interpretation segmentation method as well as the low accuracy and a lot of background noise of the existing remote sensing image segmentation based on deep learning in complex scenes, an image segmentation algorithm based on improved U-net network architecture and fully connected conditional random field was proposed. Firstly, a new network model was constructed by integrating VGG16 and U-net to effectively extract the features of remote sensing images with highly complex background. Then, by selecting the appropriate activation function and convolution method, the image segmentation accuracy was improved while the model prediction time was significantly reduced. Finally, on the basis of guaranteeing the segmentation accuracy, the segmentation result was further improved by using fully connected conditional random field. The simulation test on the standard dataset Potsdam provided by ISPRS showed that the accuracy, recall and the Mean Intersection over Union (MIoU) of the proposed algorithm were increased by 15.06 percentage points, 29.11 percentage points and 0.3662 respectively, and the Mean Absolute Error (MAE) of the algorithm was reduced by 0.02892 compared with those of U-net. Experimental results verify that the proposed algorithm is an effective and robust algorithm for extracting surface objects from remote sensing images.

Key words: deep learning, Convolutional Neural Network (CNN), depth separable convolution, fully connected conditional random field

中图分类号:

TP391.4

许玥, 冯梦如, 皮家甜, 陈勇. 基于深度学习模型的遥感图像分割方法[J]. 计算机应用, 2019, 39(10): 2905-2914.

XU Yue, FENG Mengru, PI Jiatian, CHEN Yong. Remote sensing image segmentation method based on deep learning model[J]. Journal of Computer Applications, 2019, 39(10): 2905-2914.

参考文献

[1] 高海燕, 吴波. 结合像元形状特征分割的高分辨率影像面向对象分类[J]. 遥感信息, 2010(6):67-72. (GAO H Y, WU B. Object-oriented classification of high spatial resolution remote sensing imagery based on image segmentation with pixel shape feature[J]. Remote Sensing Information, 2010(6):67-72.)
[2] 巫兆聪, 胡忠文, 张谦, 等. 结合光谱、纹理与形状结构信息的遥感影像分割方法[J]. 测绘学报, 2013, 42(1):44-50. (WU Z C, HU Z W, ZHANG Q, et al. On combining spectral, textural and shape features for remote sensing image segmentation[J]. Acta Geodaetica et Cartographica Sinica, 2013, 42(1):44-50.)
[3] YUAN J, WANG D, LI R. Remote sensing image segmentation by combining spectral and texture features[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 52(1):16-24.
[4] PALENICHKA R, DOYON F, LAKHSSASSI A, et al. Multi-scale segmentation of forest areas and tree detection in LiDAR images by the attentive vision method[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2013, 6(3):1313-1323.
[5] 都伟冰, 王双亭, 王春来. 基于机载LiDAR粗糙度指数和回波强度的道路提取[J]. 测绘科学技术学报, 2013, 30(1):63-67. (DU W B, WANG S T, WANG C L. Road extraction based on roughness index and echo intensity of airborne LiDAR[J]. Journal of Geomatics Science and Technology, 2013, 30(1):63-67.)
[6] 张曦, 胡根生, 梁栋, 等. 基于时频特征的高分辨率遥感图像道路提取[J]. 地理空间信息, 2016, 14(6):18-21, 24. (ZHANG X, HU G S, LIANG D, et al. Road extraction from high resolution remote sensing image based on time frequency feature[J]. Geospatial Information, 2016, 14(6):18-21, 24.)
[7] 周绍光, 陈超, 赫春晓. 基于形状先验和Graph Cuts原理的道路分割新方法[J]. 测绘通报, 2013(12):55-57. (ZHOU S G, CHEN C, HE C X. A new road segmentation based on shape prior and graph cuts[J]. Bulletin of Surveying and Mapping, 2013(12):55-57.)
[8] 周家香, 周安发, 陶超, 等. 一种高分辨率遥感影像城区道路网提取方法[J]. 中南大学学报(自然科学版), 2013, 44(6):2385-2391. (ZHOU J X, ZHOU A F, TAO C, et al. A methodology for urban roads network extraction from high resolution remote sensing imagery[J]. Journal of Central South University (Science & Technology), 2013, 44(6):2385-2391.)
[9] 曾发明, 杨波, 吴德文, 等. 基于Canny边缘检测算子的矿区道路提取[J]. 国土资源遥感, 2013, 25(4):72-78. (ZENG F M, YANG B, WU D W, et al. Extraction of roads in mining area based on Canny edge detection operator[J]. Remote Sensing for Land & Resources, 2013, 25(4):72-78.)
[10] BEUMIER C, IDRISSA M. Building change detection from uniform regions[C]//Proceeddings of the 2012 Iberoamerican Congress on Pattern Recognition, LNCS 7441. Berlin:Springer, 2012:648-655.
[11] HUANG X, ZHANG L, ZHU 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 & Remote Sensing, 2013, 7(1):105-115.
[12] 李炜明, 吴毅红, 胡占义. 视角和光照显著变化时的变化检测方法研究[J]. 自动化学报, 2009, 35(5):449-461. (LI W M, WU Y H, HU Z Y. Urban change detection under large view and illumination variations[J]. Acta Automatica Sinica, 2009, 35(5):449-461.)
[13] 田昊, 杨剑, 汪彦明, 等. 基于先验形状约束水平集模型的建筑物提取方法[J]. 自动化学报, 2010, 36(11):1502-1511. (TIAN H, YANG J, WANG Y M, et al. Towards automatic building extraction:variational level set model using prior shape knowledge[J]. Acta Automatica Sinica, 2010, 36(11):1502-1511.)
[14] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11):2278-2324.
[15] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems. New York:Curran Associates Inc., 2012:1097-1105.
[16] HE K, GKIOXARI G, DOLLÁR P, et al. Mask R-CNN[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway:IEEE, 2017:2961-2969.
[17] LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4):640-651.
[18] FARFADE S S, SABERIAN M J, LI L. Multi-view face detection using deep convolutional neural networks[C]//Proceedings of the 5th ACM International Conference on Multimedia Retrieval. New York:ACM, 2015:643-650.
[19] RONNEBERGER O, FISCHER P, BROX T. U-net:convolutional networks for biomedical image segmentation[C]//MICCAI 2015:Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Berlin:Springer, 2015:234-241.
[20] GLOROT X, BORDES A, BENGIO Y. Deep sparse rectifier neural networks[J]. Journal of Machine Learning Research, 2011, 15:315-323.
[21] CLEVERT D, UNTERTHINER T, HOCHREITER S. Fast and accurate deep network learning by Exponential Linear Units (ELUs)[EB/OL].[2019-01-10]. http://de.arxiv.org/pdf/1511.07289.
[22] CHOLLET F. Xception:deep learning with depthwise separable convolutions[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, 2017:1800-1807.
[23] HOWARD A G, ZHU M, CHEN B, et al. MobileNets:efficient convolutional neural networks for mobile vision applications[EB/OL].[2019-01-10]. https://arxiv.org/pdf/1704.04861.pdf.
[24] KRÄHENBVHL P, KOLTUN V. Efficient inference in fully connected CRFs with Gaussian edge potentials[C]//Proceedings of the 2011 International Conference on Neural Information Processing Systems. New York:Curran Associates Inc., 2011:109-117.
[25] ALTMAIER A, KANY C. Digital surface model generation from CORONA satellite images[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2002, 56(4):221-235.
[26] 许玥. 基于改进Unet的遥感影像语义分割在地表水体变迁中的应用[D].重庆:重庆师范大学,2019:16-35. (XU Y. Application of semantic segmentation of remote sensing image based on improved unet in surface water change[D]. Chongqing:Chongqing Normal University,2019:16-35.)
[27] ZHENG S, JAYASUMANA S, ROMERA-PAREDES B, et al. Conditional random fields as recurrent neural networks[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway:IEEE, 2015:1529-1537.
[28] GLOROT X, BENGIO Y. Understanding the difficulty of training deep feedforward neural networks[J]. Journal of Machine Learning Research, 2010, 9:249-256.
[29] ODENA A. Semi-supervised learning with generative adversarial networks[EB/OL].[2019-01-10]. https://arxiv.org/pdf/1606.01583.pdf.
[30] HUANG B, ZHAO B, SONG Y. Urban land-use mapping using a deep convolutional neural network with high spatial resolution multispectral remote sensing imagery[J]. Remote Sensing of Environment, 2018, 214:73-86.
[31] VOLPI M, TUIA D. Dense semantic labeling of subdecimeter resolution images with convolutional neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 55(2):881-893.
[32] SHERRAH J. Fully convolutional networks for dense semantic labelling of high-resolution aerial imagery[EB/OL].[2019-01-10]. https://arxiv.org/pdf/1606.02585.pdf.