Improved progressive morphological filter for digital terrain model generation from airborne lidar data

Area of Science:

• Geospatial Science
• Remote Sensing Technology
• Computational Geometry

Background:

• Airborne lidar point cloud filtering is crucial for accurate terrain mapping.
• Traditional mathematical morphology filters are efficient but struggle with terrain detail preservation.
• Existing methods often require manual parameter tuning, limiting their applicability.

Purpose of the Study:

• To develop an advanced filtering algorithm for airborne lidar point clouds in complex terrains.
• To improve upon classical progressive morphological filters by enhancing detail preservation and automating threshold selection.
• To evaluate the proposed Progressive Morphological filter with Hierarchical Radial Basis Function Interpolation (PMHR) against established benchmarks.

Main Methods:

• Refinement of the classical progressive morphological filter using hierarchical radial basis function interpolation.
• Implementation of automatic self-adaptive threshold setting for adaptive filtering.
• Inclusion of a terrain detail preservation mechanism within the filtering process.
• Validation using benchmark datasets from the International Society for Photogrammetry and Remote Sensing.

Main Results:

• The PMHR algorithm demonstrated robust performance across varied terrain features.
• Achieved an average total error of 4.27% in filtering airborne lidar point clouds.
• Attained an average Kappa coefficient of 84.57%, indicating high classification accuracy.
• Successfully preserved crucial terrain details often lost in traditional methods.

Conclusions:

• The proposed PMHR filter offers a significant improvement over classical methods for airborne lidar point cloud filtering.
• PMHR effectively balances filtering accuracy with the preservation of fine terrain features.
• The automated thresholding and detail preservation make PMHR a valuable tool for complex environmental mapping.