Gaussian Mixture Model Classification

Summary

The Gaussian mixture model uses a salient feature-based approach (Ma et al.,2016) for identifying the nonphotosynthetic components in TLS data for forest applications. This method can be used for classifying leaves, trunks, branches and ground points.

Usage

Navigate to TLS Forest > Gaussian Mixture Model Classification.

Settings

• Input Data: The input file can be a single point cloud data file or multiple data files. Point cloud data should be opened in LiDAR360 before being processed.
• Sub-Group Information: Select the sample data for trunk, leaves and ground for model calibration (this can be done using the select tools to select different samples from the TLS point cloud), in the "Sub-Group Information" area, indication the scale and load the sample data in the "Sample File Information" area (each sub-group type needs to be highlighted individually and the appropriate sample data loaded from within the Sample File Information area). Ideal scale parameters can be estimated using the guide below:
  • Trunk: This includes the diameter of trunk and branches. Branches and twigs can be rather small, and users should select the minimum number that is appropriate for small branches. Note: Too small numbers may also cause misclassification.
  • Leaves: The width or length of leaves. Some leaves can be very small, so the selected values should be small enough to include these leaves.
  • Ground: The spacing of ground points.

Below is a figure showing the result after performing the Gaussian Mixture Model Classification.

    @inproceedings{
        author={Ma L X, Zheng G, Eitel J U H, Moskal L M, He W and Huang H},
        title={Improved salient feature-based approach for automatically separating photosynthetic and nonphotosynthetic components within terrestrial lidar point cloud data of forest canopies},
        booktitle={IEEE Transactions on Geoscience and Remote Sensing, 54(2): 679-696},
        year={2016}
    }