MaxEnt: A Maximum Entropy-Based Optimal Neighbor Selection for Multispectral Airborne LiDAR Point Cloud Classification

Paper

Introduction

The idea came from a fairly practical question: when you have a point with both elevation and multispectral intensity (532 nm, 1064 nm, 1550 nm), how do you decide which neighbor points should be used to extract features for classification?

Most methods use a fixed neighborhood — spherical, cylindrical, or k-nearest. That works, but near class boundaries it almost always pulls in points from other classes. That noise then propagates into the features. So we tried something else: let the data itself decide. For each point, MaxEnt looks at the initial neighbors, measures the "information difference" in elevation and intensity, and finds the threshold that maximizes entropy — splitting the neighborhood into homogeneous (keep) and heterogeneous (discard) sets.

Citation

If you find our work useful for your research, please cite:

@article{jiang2023maximum,
  title={A maximum entropy-based optimal neighbor selection for multispectral airborne LiDAR point cloud classification},
  author={Jiang, Ge and Yan, Wai Yeung and Lichti, Derek D},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  volume={61},
  pages={1--18},
  year={2023},
  publisher={IEEE}
}