Forestry Identification With Lidar Waveform and Point Clouds

Abstract

The aim of this study was to analyze discrete and waveform data to improve existing Terrain Classification (TERCAT) capabilities. Light Detection and Ranging (LiDAR) data were collected over the Point Lobos State Park, which contains various buildings, vegetation, and man-made surfaces. Data were used from two separate airborne LiDAR systems, Optech Titan and Airborne Hydrography AB (AHAB) Chiroptera II. Classic standard point cloud analysis techniques were used with the discrete data. Waveform data were analyzed following a gridding or rasterization process to enable visualization and processing. Analysis approaches used were ENVI classification tools such as Support Vector Machines (SVM), Spectral Angle Mapper (SAM), Maximum Likelihood, and K-means to classify returns. Through the use of this analog to hyperspectral data analysis to classify vegetation and terrain, the results are that, by using the Support Vector Machines with full waveform data, we can successfully improve low vegetation classifiers by 40 , and differentiate tree types (Pine/Cypress) at 40-60% accuracy.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 2018
Accession Number
AD1059826

Entities

People

  • Andrew S. Davis

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Autonomy
  • Energy and Power Technologies
  • Materials and Manufacturing Processes
  • Sensors
  • Space

DTIC Thesaurus Topics

  • Accuracy
  • Data Mining
  • Detection
  • Detectors
  • Dimensionality Reduction
  • Earth Sciences
  • Forestry
  • Forests
  • Geography
  • Information Science
  • Lidar
  • Machine Learning
  • Point Clouds
  • Remote Sensing
  • Supervised Machine Learning
  • Synthetic Aperture Radar
  • Unsupervised Machine Learning

Readers

  • Geodesy
  • Military/Explosive Ordnance Disposal (EOD) Technology
  • Oceanography.

Technology Areas

  • AI & ML