Fully Convolutional Networks and Geographic Object-Based Image Analysis for the Classification of VHR Imagery

被引:54
|
作者
Mboga, Nicholus [1 ]
Georganos, Stefanos [1 ]
Grippa, Tais [1 ]
Lennert, Moritz [1 ]
Vanhuysse, Sabine [1 ]
Wolff, Eleonore [1 ]
机构
[1] ULB, Dept Geosci Environm & Soc, B-1050 Brussels, Belgium
关键词
fully convolutional networks; convolutional neural networks; remote sensing; very high resolution; landcover classification; geographical object-based image analysis; NEURAL-NETWORKS; SELECTION;
D O I
10.3390/rs11050597
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Land cover Classified maps obtained from deep learning methods such as Convolutional neural networks (CNNs) and fully convolutional networks (FCNs) usually have high classification accuracy but with the detailed structures of objects lost or smoothed. In this work, we develop a methodology based on fully convolutional networks (FCN) that is trained in an end-to-end fashion using aerial RGB images only as input. Skip connections are introduced into the FCN architecture to recover high spatial details from the lower convolutional layers. The experiments are conducted on the city of Goma in the Democratic Republic of Congo. We compare the results to a state-of-the art approach based on a semi-automatic Geographic object image-based analysis (GEOBIA) processing chain. State-of-the art classification accuracies are obtained by both methods whereby FCN and the best baseline method have an overall accuracy of 91.3% and 89.5% respectively. The maps have good visual quality and the use of an FCN skip architecture minimizes the rounded edges that is characteristic of FCN maps. Additional experiments are done to refine FCN classified maps using segments obtained from GEOBIA generated at different scale and minimum segment size. High OA of up to 91.5% is achieved accompanied with an improved edge delineation in the FCN maps, and future work will involve explicitly incorporating boundary information from the GEOBIA segmentation into the FCN pipeline in an end-to-end fashion. Finally, we observe that FCN has a lower computational cost than the standard patch-based CNN approach especially at inference.
引用
收藏
页数:17
相关论文
共 50 条
  • [1] AN OBJECT-BASED APPROACH TO VHR IMAGE CLASSIFICATION
    Asma, Semcheddine Belkis
    Abdelhamid, Daamouche
    [J]. 2020 MEDITERRANEAN AND MIDDLE-EAST GEOSCIENCE AND REMOTE SENSING SYMPOSIUM (M2GARSS), 2020, : 93 - 96
  • [2] Object-based classification for mangrove with VHR remotely sensed image
    Liu, Zhigang
    Li, Jing
    Lim, Boonleong
    Seng, Chungyueh
    Inbaraj, Suppiah
    [J]. GEOINFORMATICS 2007: REMOTELY SENSED DATA AND INFORMATION, PTS 1 AND 2, 2007, 6752
  • [3] RETRACTED: Object-based Image Analysis of VHR Imagery by Combining Spectrum and Texture (Retracted Article)
    Liu, Yue
    Chen, Jianming
    Wu, Guangmin
    Chu, Hequn
    [J]. 2011 INTERNATIONAL CONFERENCE ON ENERGY AND ENVIRONMENTAL SCIENCE-ICEES 2011, 2011, 11
  • [4] OBJECT-BASED CHANGE DETECTION MODEL USING CORRELATION ANALYSIS AND CLASSIFICATION FOR VHR IMAGE
    Tang, Zhipeng
    Tang, Hong
    He, Shi
    Mao, Ting
    [J]. 2015 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM (IGARSS), 2015, : 4840 - 4843
  • [5] A Noise Removal Approach for Object-based Classification of VHR Imagery via Post-classification
    Zheng, Laiwen
    Wan, Lihong
    Huo, Hong
    Fang, Tao
    [J]. 2014 INTERNATIONAL CONFERENCE ON AUDIO, LANGUAGE AND IMAGE PROCESSING (ICALIP), VOLS 1-2, 2014, : 915 - 920
  • [6] Thematic accuracy assessment of geographic object-based image classification
    Radoux, J.
    Bogaert, P.
    Fasbender, D.
    Defourny, P.
    [J]. INTERNATIONAL JOURNAL OF GEOGRAPHICAL INFORMATION SCIENCE, 2011, 25 (06) : 895 - 911
  • [7] Fuzzy segmentation for geographic object-based image analysis
    Lizarazo, Ivan
    Elsner, Paul
    [J]. REMOTE SENSING FOR ENVIRONMENTAL MONITORING, GIS APPLICATIONS, AND GEOLOGY IX, 2009, 7478
  • [8] Object-Based Classification of Urban Areas Using VHR Imagery and Height Points Ancillary Data
    Salehi, Bahram
    Zhang, Yun
    Zhong, Ming
    Dey, Vivek
    [J]. REMOTE SENSING, 2012, 4 (08) : 2256 - 2276
  • [9] Recurrent Multiresolution Convolutional Networks for VHR Image Classification
    Bergado, John Ray
    Persello, Claudio
    Stein, Alfred
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2018, 56 (11): : 6361 - 6374
  • [10] Classification of Soil Types Using Geographic Object-Based Image Analysis and Random Forests
    Andrei DORNIK
    Lucian DRAGUT
    Petru URDEA
    [J]. Pedosphere, 2018, (06) : 913 - 925