High-resolution remote sensing image semantic segmentation based on semi-supervised full convolution network method

被引：0

作者：

Geng Y. ^{[1
,2
]}

Tao C. ^{[1
,2
]}

Shen J. ^{[1
,2
]}

Zou Z. ^{[1
,2
]}

机构：

[1] School of Geosciences and Info-Physics, Central South University, Changsha

[2] Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring, Central South University, Ministry of Education, Changsha

来源：

Cehui Xuebao/Acta Geodaetica et Cartographica Sinica | 2020年 / 49卷 / 04期

基金：

中国国家自然科学基金;

关键词：

Full convolution network; Remote sensing image; Semantic segmentation; Semi-supervised;

D O I：

10.11947/j.AGCS.2020.20190044

中图分类号：

学科分类号：

摘要：

In the field of remote sensing, the method of realizing image semantic segmentation by using a large amount of label image data to supervise training full convolution network will result in expensive label drawing cost, while the use of a small amount of label data would lead to network performance degradation. To solve this problem, this paper proposes a semi-supervised full convolution network based semantic segmentation method for high resolution remote sensing images. Specifically, we explore an ensemble prediction technique to train the end-to-end semantic segmentation network by simultaneously optimizing a standard supervised classification loss on labeled samples along with an additional unsupervised consistence loss term imposed on labeled and unlabeled data. In the experiments, the image data set of Vaihingen in Germany provided by ISPRS and satellite GF-1 data were used, and the experimental results show that the proposed method can effectively improve the network performance degradation caused by using only a small amount of label data. © 2020, Surveying and Mapping Press. All right reserved.

引用

页码：499 / 508

页数：9

共 23 条

[1] He X., Zou Z., Tao C., Et al., Combined saliency with multi-convolutional neural network for high resolution remote sensing scene classification, Acta Geodaetica et Cartographica Sinica, 45, 9, pp. 1073-1080, (2016)
[2] Zuo T., Feng J., Chen X., HF-FCN: hierarchically fused fully convolutional network for robust building extraction, Proceedings of the 13th Asian Conference on Computer Vision, (2016)
[3] Noronha S., Nevatia R., Detection and modeling of buildings from multiple aerial images, IEEE Transactions on Pattern Analysis and Machine Intelligence, 23, 5, pp. 501-518, (2001)
[4] Cote M., Saeedi P., Automatic rooftop extraction in nadir aerial imagery of suburban regions using corners and variational level set evolution, IEEE Transactions on Geoscience and Remote Sensing, 51, 1, pp. 313-328, (2013)
[5] Li E., Femiani J., Xu S., Et al., Robust rooftop extraction from visible band images using higher order CRF, IEEE Transactions on Geoscience and Remote Sensing, 53, 8, pp. 4483-4495, (2015)
[6] Hu X., Gong X., Zhang M., A variational approach for automatic man-made object detection from remote sensing images, Acta Geodaetica et Cartographica Sinica, 47, 6, pp. 780-789, (2018)
[7] Lin X., Zhang J., Object-based morphological building index for building extraction from high resolution remote sensing imagery, Acta Geodaetica et Cartographica Sinica, 46, 6, pp. 724-733, (2017)
[8] Wang W., Yang N., Zhang Y., Et al., A review of road extraction from remote sensing images, Journal of Traffic and Transportation Engineering, 3, 3, pp. 271-282, (2016)
[9] Lecun Y., Bottou L., Bengio Y., Et al., Gradient-based learning applied to document recognition, Proceedings of the IEEE, 86, 11, pp. 2278-2324, (1998)
[10] Krizhevsky A., Sutskever I., Hinton G.E., ImageNet classification with deep convolutional neural networks, Proceedings of the 25th International Conference on Neural Information Processing Systems, (2012)

← 1 2 3 →