SAR-Colorization

Code for Paper Q. Song, F. Xu, and Y.Q. Jin, "Radar Image Colorization: Converting Single-Polarization to Fully Polarimetric Using Deep Neural Networks," IEEE Access.

• Datasets that used in the paper can be downloaded from https://www.asf.alaska.edu/ select UAVSAR data acquired at San Andreas fault.

Prerequisites

• Python 2.7
• Tensorflow
• SciPy
• NumPy
• h5py

Useage

First, download dataset:

$ Download the data(https://www.asf.alaska.edu/), and save it as *.mat* file. Two Images used in this paper are SanAnd_08503_12130_002_121109_L090_CX_01 and gulfco_00702_12053_004_120702_L090_CX_01.
$  mkdir data
 ... and copy data into it.

To train a model:

$ python main.py

Results

VV-pol SAR Image

Actual Full-pol SAR Image

Reconstructed Full-pol Image

Authors

• Qian Song
• Feng Xu

Reference

[1] Q. Song, F. Xu, and Y.Q. Jin, "Radar Image Colorization: Converting Single-Polarization to Fully Polarimetric Using Deep Neural Networks," IEEE Access.

[2] G. Larsson, M. Maire, and G. Shakhnarovich, “Learning representations for automatic colorization,” arXiv: 1603.06668, 2016.