A Semantic Segmentation Approach Based on DeepLab Network in High-Resolution Remote Sensing Images

Recently, more and more applications for high-resolution remote sensing image intelligent processing are required. Therefore, the semantic segmentation based on deep learning has successfully attracted people’s attention. In this paper, the improved Deeplabv3 network is used in the application of image semantic segmentation. The problem of segmenting objects of multiple scales of high-resolution remote sensing image is handled, and the Chinese GaoFen NO. 2(GF-2) remote sensing image is taken as the main research object. Firstly, the original image is pre-processed. Next, use data augmentation and expansion for the pre-processed training image to avoid over-fitting. Finally, it is studied the adaptability and accuracy of the model of high-resolution remote sensing images, while is found the appropriate parameters to improve the precise of the result models compared. And explore the effectiveness of the model in the case of a fewer samples. This model is demonstrated that could be achieved the well classification result.

[1]  Weizheng Wang,et al.  Development of convolutional neural network and its application in image classification: a survey , 2019, Optical Engineering.

[2]  Chokri Ben Amar,et al.  Deep learning for semantic segmentation of remote sensing images with rich spectral content , 2017, 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).

[3]  Ronald Kemker,et al.  Algorithms for semantic segmentation of multispectral remote sensing imagery using deep learning , 2017, ISPRS Journal of Photogrammetry and Remote Sensing.

[4]  Jian Sun,et al.  Identity Mappings in Deep Residual Networks , 2016, ECCV.

[5]  Iasonas Kokkinos,et al.  DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs , 2016, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[6]  Xiaodong Mu,et al.  Scene classification of remote sensing image based on deep network grading transferring , 2018, Optik.

[7]  Vladlen Koltun,et al.  Multi-Scale Context Aggregation by Dilated Convolutions , 2015, ICLR.

[8]  Ian D. Reid,et al.  RefineNet: Multi-path Refinement Networks for High-Resolution Semantic Segmentation , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[9]  Menglong Yan,et al.  A new semantic segmentation model for remote sensing images , 2017, 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).

[10]  Abhishek Samanta,et al.  A Review of Convolutional Neural Networks , 2020, 2020 International Conference on Emerging Trends in Information Technology and Engineering (ic-ETITE).

[11]  George Papandreou,et al.  Rethinking Atrous Convolution for Semantic Image Segmentation , 2017, ArXiv.

[12]  Zhenwei Shi,et al.  MugNet: Deep learning for hyperspectral image classification using limited samples , 2017, ISPRS Journal of Photogrammetry and Remote Sensing.

[13]  Iasonas Kokkinos,et al.  Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs , 2014, ICLR.

[14]  Xiaoyu Chen,et al.  A review on image classification of remote sensing using deep learning , 2017, 2017 3rd IEEE International Conference on Computer and Communications (ICCC).

[15]  Xiaogang Wang,et al.  Pyramid Scene Parsing Network , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[16]  José García Rodríguez,et al.  A Review on Deep Learning Techniques Applied to Semantic Segmentation , 2017, ArXiv.

[17]  Zhang Jiaxing,et al.  Combined Saliency with Multi-Convolutional Neural Network for High Resolution Remote Sensing Scene Classification , 2016 .

[18]  Trevor Darrell,et al.  Fully Convolutional Networks for Semantic Segmentation , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.