Reconstruction Full-Pol SAR Data from Single-Pol SAR Image Using Deep Neural Network

Compared with single channel polarimetric (single-pol) SAR image, full polarimetric (full-pol) data convey richer information, but with compromises on higher system complexity and lower resolution or swath. In order to balance these factors, a deep neural networks based method is proposed to recover full-pol data from single-pol data in this paper. It consists of two parts:a feature extractor network is applied first to extract hierarchical multi-scale spatial features, followed by a feature translator network to predict polarimetric features with which full-pol SAR data can be recovered. Both qualitative and quantitative results show that the recovered full-pol SAR data agrees well with the real full-pol data. No prior information is assumed for scatterer media, and the framework can be easily expanded to recovery full-pol data from non-full-pol data. Traditional PolSAR applications such as model-based decomposition and unsupervised classification can now be applied directly to recovered full-pol SAR image to interpret the physical scattering mechanism.

[1]  Ya-Qiu Jin,et al.  Deorientation theory of polarimetric scattering targets and application to terrain surface classification , 2005, IEEE Trans. Geosci. Remote. Sens..

[2]  Ian McLoughlin,et al.  Neural network-assisted reconstruction of full polarimetric SAR information , 2010, 2010 4th International Symposium on Communications, Control and Signal Processing (ISCCSP).

[3]  Thomas L. Ainsworth,et al.  Unsupervised classification of polarimetric synthetic aperture Radar images using fuzzy clustering and EM clustering , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[4]  Stephen L. Durden,et al.  A three-component scattering model for polarimetric SAR data , 1998, IEEE Trans. Geosci. Remote. Sens..

[5]  Jean-Claude Souyris,et al.  Compact polarimetry based on symmetry properties of geophysical media: the /spl pi//4 mode , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[6]  Gregory Shakhnarovich,et al.  Learning Representations for Automatic Colorization , 2016, ECCV.

[7]  Eric Pottier,et al.  An entropy based classification scheme for land applications of polarimetric SAR , 1997, IEEE Trans. Geosci. Remote. Sens..

[8]  Dong-Xiao Yue,et al.  Wishart–Bayesian Reconstruction of Quad-Pol From Compact-Pol SAR Image , 2017, IEEE Geoscience and Remote Sensing Letters.

[9]  Jean-Claude Souyris,et al.  Polarimetry based on one transmitting and two receiving polarizations: the /spl pi//4 mode , 2002, IEEE International Geoscience and Remote Sensing Symposium.

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

[11]  Qian Song,et al.  Polarimetric SAR Image Factorization , 2017, IEEE Transactions on Geoscience and Remote Sensing.

[12]  Qian Song,et al.  Radar Image Colorization: Converting Single-Polarization to Fully Polarimetric Using Deep Neural Networks , 2017, IEEE Access.

[13]  R. Keith Raney,et al.  Hybrid-Polarity SAR Architecture , 2006, 2006 IEEE International Symposium on Geoscience and Remote Sensing.

[14]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.