Classification of very high resolution SAR image based on convolutional neural network

The new advanced very high resolution (VHR) synthetic aperture radar (SAR) sensor is a kind of high-tech imaging radar developed rapidly in recent years, and it can get even less than 1 m high resolution SAR image. The feature of the VHR SAR image is different from the low or medium resolution SAR image and it contains more abundant information, so the traditional SAR image classification methods can't be directly applied in VHR SAR image classification. In order to achieve high precision classification performance of the VHR SAR image, convolutional neural network (CNN), a kind of representative deep learning method, is applied in this paper. Compared with the traditional supervised classification methods, such as minimum distance and maximum likelihood, the CNN method obtained better classification result with 97.0% average accuracy. The experiments demonstrate that the CNN is an effective and favorable classification method for VHR SAR image classification.

[1]  Xiaogang Wang,et al.  Hybrid Deep Learning for Face Verification , 2013, 2013 IEEE International Conference on Computer Vision.

[2]  Jie Geng,et al.  High-Resolution SAR Image Classification via Deep Convolutional Autoencoders , 2015, IEEE Geoscience and Remote Sensing Letters.

[3]  Christian Cochin,et al.  Classification of ships using real and simulated data in a convolutional neural network , 2016, 2016 IEEE Radar Conference (RadarConf).

[4]  Jian Yang,et al.  Application of deep learning to polarimetric SAR classification , 2015 .

[5]  David A. Clausi,et al.  Sea Ice Concentration Estimation During Melt From Dual-Pol SAR Scenes Using Deep Convolutional Neural Networks: A Case Study , 2016, IEEE Transactions on Geoscience and Remote Sensing.

[6]  Trevor Darrell,et al.  Caffe: Convolutional Architecture for Fast Feature Embedding , 2014, ACM Multimedia.

[7]  Yi Su,et al.  Unsupervised polarimetric SAR urban area classification based on model-based decomposition with cross scattering , 2016 .

[8]  Jin Zhao,et al.  Discriminant deep belief network for high-resolution SAR image classification , 2017, Pattern Recognit..

[9]  Xiaogang Wang,et al.  Deep Learning Face Representation by Joint Identification-Verification , 2014, NIPS.

[10]  Shuang Wang,et al.  Polarimetric SAR images classification using deep belief networks with learning features , 2015, 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).

[11]  Xiaogang Wang,et al.  Hybrid Deep Learning for Face Verification , 2013, ICCV.

[12]  Domenico Velotto,et al.  Target classification in oceanographic SAR images with deep neural networks: Architecture and initial results , 2015, 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).

[13]  Fang Liu,et al.  POL-SAR Image Classification Based on Wishart DBN and Local Spatial Information , 2016, IEEE Transactions on Geoscience and Remote Sensing.

[14]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[15]  Alexander Kaptein,et al.  From TerraSAR-X towards TerraSAR Next Generation , 2014 .

[16]  Haipeng Wang,et al.  Target Classification Using the Deep Convolutional Networks for SAR Images , 2016, IEEE Transactions on Geoscience and Remote Sensing.

[17]  Jürgen Janoth,et al.  Terrasar next generation - Mission capabilities , 2013, 2013 IEEE International Geoscience and Remote Sensing Symposium - IGARSS.

[18]  Shuang Wang,et al.  Wishart RBM based DBN for polarimetric synthetic radar data classification , 2015, 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).