Message Board

Respected readers, authors and reviewers, you can add comments to this page on any questions about the contribution, review,        editing and publication of this journal. We will give you an answer as soon as possible. Thank you for your support!

Name
E-mail
Phone
Title
Content
Verification Code
Turn off MathJax
Article Contents

HE Xiaohui, CHEN Mingyang, LI Panle, TIAN Zhihui, ZHOU Guangsheng. Road Extraction from Remote Sensing Image by Integrating DCNN with Short Range Conditional Random Field(SRCRF)[J]. Geomatics and Information Science of Wuhan University. doi: 10.13203/j.whugis20210464
Citation: HE Xiaohui, CHEN Mingyang, LI Panle, TIAN Zhihui, ZHOU Guangsheng. Road Extraction from Remote Sensing Image by Integrating DCNN with Short Range Conditional Random Field(SRCRF)[J]. Geomatics and Information Science of Wuhan University. doi: 10.13203/j.whugis20210464

Road Extraction from Remote Sensing Image by Integrating DCNN with Short Range Conditional Random Field(SRCRF)

doi: 10.13203/j.whugis20210464
Funds:

The Second Comprehensive Investigation and Research Project on Qinghai-Tibet Plateau.(2019QZKK0106)

  • Received Date: 2022-06-24
    Available Online: 2022-08-10
  • Objective: Deep Convolutional Neural Network (DCNN) is widely used in automatic road extraction from high-resolution Remote Sensing images (HRSIs). However, the existing methods are difficult to model the context relationship between pixels in the predicted results. To solve this problem, some studies have used Fully Connected Crf (FullCrf) to perform secondary optimization of semantic segmentation results combined with context information, but the discontinuity problem of road structure cannot be effectively improved. In order to improve the integrity of road structure, this study proposes a Short Range Conditional Random Filed (SRCRF) model combined with DCNN. Methods: SRCRF mainly includes unary potential function based on road pre-segmentation, binary potential function based on spectral spatial features and k-neighborhood mean field inference algorithm. Firstly, the priori knowledge of road pre-segmentation results is obtained by using the powerful feature extraction capability of DCNN as the unary potential function of SRCRF, Secondly, the dependence of the binary potential function defined by the linear combination of Gaussian kernel functions on the surrounding nodes is modeled. The binary potential function enables the classification results to have local consistency that is, adjacent pixels with similar spectral features have the same label. Finally, K-neighborhood mean field inference algorithm based on mean field approximation inference algorithm optimizes the inference range to make full use of the spatial context information and spectral feature context information of the road, and then calculates the optimal label corresponding to each pixel based on the space and spectral feature to optimize the road accurately. The convolution method is adopted to control the inference range of SRCRF within the radius of K in order to improve the proportion of feature vectors (road-road). Results: The experimental results show that SRCRF alleviates the transition smoothness of FullCrf, and alleviates the structural discontinuity and incompleteness in the road acquisition results of high resolution remote sensing images. In Zimbawe-Roads dataset and Cheng-Roads dataset, F1 values of SRCRF increased by about 4.01% and 3.73% respectively compared with DCNN, and about 3.25% and 2.28% respectively compared with FullCrf. Conclusion: Compared with traditional deep learning methods for road extraction from high resolution remote sensing images, this paper proposes a new road extraction scheme for remote sensing images, SRCRF. This scheme combines the advantages of DCNN and optimizes the fully connected structure of traditional conditional random fields into k-neighborhood structure, which reduces the inference scope and improves the proportion of feature vectors (road to road). Compared with FullCrf, SRCRF can make better use of image color features and spatial features to accurately optimize the road extraction results of deep learning output. According to the results of SRCRF on Cheng-Roads and Zimbawe-Roads data sets, the performance of this method is improved compared with DCNN and FullCrf, and the time is shortened by one order of magnitude compared with FullCrf. In future work, we will further investigate the potential for learning Gaussian features and investigate more complex CRF architectures to better capture global context information. Finally, we are particularly interested in exploring the application potential of SRCRF in other fields, such as building extraction, vehicle extraction, lake extraction, etc.
  • [1] Abdollahi A, Pradhan B, Shukla N, et al.Deep Learning Approaches Applied to Remote Sensing Datasets for Road Extraction:A State-of-the-Art Review[J].Remote Sensing, 2020, 12(9):1444
    [2] Soulard C E, Acevedo W, Stehman S V.Removing Rural Roads from the National Land Cover Database to Create Improved Urban Maps for the United States, 1992 to 2011[J].Photogrammetric Engineering&Remote Sensing, 2018, 84(2):101-109
    [3] Li D M, Deng L B, Cai Z M, et al.Notice of Retraction:Intelligent Transportation System in Macao Based on Deep Self-Coding Learning[J].IEEE Transactions on Industrial Informatics, 2018, 14(7):3253-3260
    [4] Vargas-Muñoz J E, Lobry S, Falcão A X, et al.Correcting Rural Building Annotations in OpenStreetMap Using Convolutional Neural Networks[J].ISPRS Journal of Photogrammetry and Remote Sensing, 2019, 147:283-293
    [5] Xu Y Y, Xie Z, Wu L, et al.Multilane Roads Extracted from the OpenStreetMap Urban Road Network Using Random Forests[J].Transactions in GIS, 2019, 23(2):224-240
    [6] Bruzzone L, Persello C.A Novel Context-Sensitive Semisupervised SVM Classifier Robust to Mislabeled Training Samples[J].IEEE Transactions on Geoscience and Remote Sensing, 2009, 47(7):2142-2154
    [7] 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
    [8] Badrinarayanan V, Kendall A, Cipolla R.SegNet:A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12):2481-2495
    [9] Abderrahim N Y Q, Abderrahim S, Rida A.Road Segmentation using U-Net architecture[C]//2020 IEEE International conference of Moroccan Geomatics (Morgeo).Casablanca, Morocco.:1-4
    [10] Ronneberger O, Fischer P, Brox T.U-Net:Convolutional Networks for Biomedical Image Segmentation[M]//Lecture Notes in Computer Science.Cham:Springer International Publishing, 2015:234-241
    [11] Cheng G L, Wang Y, Xu S B, et al.Automatic Road Detection and Centerline Extraction via Cascaded End-to-End Convolutional Neural Network[J].IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(6):3322-3337
    [12] Zhang Z X, Liu Q J, Wang Y H.Road Extraction by Deep Residual U-Net[J].IEEE Geoscience and Remote Sensing Letters, 2018, 15(5):749-753
    [13] He K M, Zhang X Y, Ren S Q, et al.Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas, NV, USA.:770-778
    [14] He H, Yang D F, Wang S C, et al.Road Extraction by Using Atrous Spatial Pyramid Pooling Integrated Encoder-Decoder Network and Structural Similarity Loss[J].Remote Sensing, 2019, 11(9):1015
    [15] Chen L C, Papandreou G, Schroff F, et al.Rethinking Atrous Convolution for Semantic Image Segmentation[EB/OL].2017:arXiv:1706.05587.https://arxiv.org/abs/1706.05587
    [16] Zhao H S, Shi J P, Qi X J, et al.Pyramid scene parsing network[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition.Honolulu, HI, USA.:6230-6239
    [17] Rao Y B, Liu W, Pu J S, et al.Roads detection of aerial image with FCN-CRF model[C]//2018 IEEE Visual Communications and Image Processing.Taichung, China.:1-4
    [18] Krähenbühl P, Koltun V.Efficient Inference in Fully Connected CRFS with Gaussian Edge Potentials[EB/OL].2012:arXiv:1210.5644.https://arxiv.org/abs/1210.5644
    [19] Alam F I, Zhou J, Liew A W C, et al.Conditional Random Field and Deep Feature Learning for Hyperspectral Image Classification[J].IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(3):1612-1628
    [20] Alam F I, Zhou J, Liew A W C, et al.CRF learning with CNN features for hyperspectral image segmentation[C]//2016 IEEE International Geoscience and Remote Sensing Symposium.Beijing, China.:6890-6893
    [21] Teichmann M, Cipolla R.Convolutional CRFS for Semantic Segmentation[J].ArXiv Preprint ArXiv:180504777, 2018
    [22] Li Q Y, Shi Y L, Huang X, et al.Building Footprint Generation by Integrating Convolution Neural Network with Feature Pairwise Conditional Random Field (FPCRF)[J].IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(11):7502-7519
    [23] Li S Z.Markov random field modeling in image analysis[M].3rd ed.London:Springer, 2009
    [24] Adams A, Baek J, Davis M A.Fast High-Dimensional Filtering Using the Permutohedral Lattice[J].Computer Graphics Forum, 2010, 29(2):753-762
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Article Metrics

Article views(74) PDF downloads(3) Cited by()

Related
Proportional views

Road Extraction from Remote Sensing Image by Integrating DCNN with Short Range Conditional Random Field(SRCRF)

doi: 10.13203/j.whugis20210464
Funds:

The Second Comprehensive Investigation and Research Project on Qinghai-Tibet Plateau.(2019QZKK0106)

Abstract: Objective: Deep Convolutional Neural Network (DCNN) is widely used in automatic road extraction from high-resolution Remote Sensing images (HRSIs). However, the existing methods are difficult to model the context relationship between pixels in the predicted results. To solve this problem, some studies have used Fully Connected Crf (FullCrf) to perform secondary optimization of semantic segmentation results combined with context information, but the discontinuity problem of road structure cannot be effectively improved. In order to improve the integrity of road structure, this study proposes a Short Range Conditional Random Filed (SRCRF) model combined with DCNN. Methods: SRCRF mainly includes unary potential function based on road pre-segmentation, binary potential function based on spectral spatial features and k-neighborhood mean field inference algorithm. Firstly, the priori knowledge of road pre-segmentation results is obtained by using the powerful feature extraction capability of DCNN as the unary potential function of SRCRF, Secondly, the dependence of the binary potential function defined by the linear combination of Gaussian kernel functions on the surrounding nodes is modeled. The binary potential function enables the classification results to have local consistency that is, adjacent pixels with similar spectral features have the same label. Finally, K-neighborhood mean field inference algorithm based on mean field approximation inference algorithm optimizes the inference range to make full use of the spatial context information and spectral feature context information of the road, and then calculates the optimal label corresponding to each pixel based on the space and spectral feature to optimize the road accurately. The convolution method is adopted to control the inference range of SRCRF within the radius of K in order to improve the proportion of feature vectors (road-road). Results: The experimental results show that SRCRF alleviates the transition smoothness of FullCrf, and alleviates the structural discontinuity and incompleteness in the road acquisition results of high resolution remote sensing images. In Zimbawe-Roads dataset and Cheng-Roads dataset, F1 values of SRCRF increased by about 4.01% and 3.73% respectively compared with DCNN, and about 3.25% and 2.28% respectively compared with FullCrf. Conclusion: Compared with traditional deep learning methods for road extraction from high resolution remote sensing images, this paper proposes a new road extraction scheme for remote sensing images, SRCRF. This scheme combines the advantages of DCNN and optimizes the fully connected structure of traditional conditional random fields into k-neighborhood structure, which reduces the inference scope and improves the proportion of feature vectors (road to road). Compared with FullCrf, SRCRF can make better use of image color features and spatial features to accurately optimize the road extraction results of deep learning output. According to the results of SRCRF on Cheng-Roads and Zimbawe-Roads data sets, the performance of this method is improved compared with DCNN and FullCrf, and the time is shortened by one order of magnitude compared with FullCrf. In future work, we will further investigate the potential for learning Gaussian features and investigate more complex CRF architectures to better capture global context information. Finally, we are particularly interested in exploring the application potential of SRCRF in other fields, such as building extraction, vehicle extraction, lake extraction, etc.

HE Xiaohui, CHEN Mingyang, LI Panle, TIAN Zhihui, ZHOU Guangsheng. Road Extraction from Remote Sensing Image by Integrating DCNN with Short Range Conditional Random Field(SRCRF)[J]. Geomatics and Information Science of Wuhan University. doi: 10.13203/j.whugis20210464
Citation: HE Xiaohui, CHEN Mingyang, LI Panle, TIAN Zhihui, ZHOU Guangsheng. Road Extraction from Remote Sensing Image by Integrating DCNN with Short Range Conditional Random Field(SRCRF)[J]. Geomatics and Information Science of Wuhan University. doi: 10.13203/j.whugis20210464
Reference (24)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return