Hierarchical-Random-Walk-network-for-Hyperspectral-and-LiDAR-classification

This example implements the paper in review [Joint Classification of Hyperspectral and LiDAR Data Using Hierarchical Random Walk and Deep CNN Architecture]

A Joint Classification method of Hyperspectral and LiDAR Data Using Hierarchical Random Walk and Deep CNN Architecture. Reach a quite high classification accuracy. Evaluated on the dataset of Houston, Trento and MUUFL.

Prerequisites

• Python 2.7 or 3.6
• Packages

pip install -r requirements.txt

Usage

Data set links

1. Houston dataset were introduced for the 2013 IEEE GRSS Data Fusion contest. Data set links comes from http://www.grss-ieee.org/community/technical-committees/data-fusion/2013-ieee-grss-data-fusion-contest/

2. The authors would like to thank Dr. P. Ghamisi for providing the Trento Data.

3. The MUUFL Gulfport Hyperspectral and LIDAR Data [1][2] is Available from https://github.com/GatorSense/MUUFLGulfport/.

[1] P. Gader, A. Zare, R. Close, J. Aitken, G. Tuell, “MUUFL Gulfport Hyperspectral and LiDAR Airborne Data Set,” University of Florida, Gainesville, FL, Tech. Rep. REP-2013-570, Oct. 2013.

[2] X. Du and A. Zare, “Technical Report: Scene Label Ground Truth Map for MUUFL Gulfport Data Set,” University of Florida, Gainesville, FL, Tech. Rep. 20170417, Apr. 2017. Available: http://ufdc.ufl.edu/IR00009711/00001.

dataset utilization

Use Gramm-Schmidt method in ENVI to merge HSI and LiDAR-based DSM

Please modify line 10-23 in data_util_c.py for the dataset details.

Training

Train the merged HSI and LiDAR-based DSM

python main.py --train merge --epochs 20 

save pred.npy and index.npy in （.mat）model

Hierarchical Random Walk Optimization

run HBRW.m in Matlab

Results

All the results are cited from original paper. More details can be found in the paper.

dataset	Kappa	OA
Houston	93.09%	93.61%
Trento	98.48%	98.86%
MUUFL	92.52%	94.31%

Citation

Please kindly cite the papers if this code is useful and helpful for your research.

Zhao X, Tao R, Li W, et al. Joint Classification of Hyperspectral and LiDAR Data Using Hierarchical Random Walk and Deep CNN Architecture[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020.

@article{zhao2020joint,
  title={Joint Classification of Hyperspectral and LiDAR Data Using Hierarchical Random Walk and Deep CNN Architecture},
  author={Zhao, Xudong and Tao, Ran and Li, Wei and Li, Heng-Chao and Du, Qian and Liao, Wenzhi and Philips, Wilfried},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  year={2020},
  publisher={IEEE}
}

TODO

1. pytorch version.
2. more flexiable dataset utilization