The benefits of specral LIDAR for FOPEN applications

Abstract

Object detection and classification in foliage penetration (FOPEN) situations is difficult due to canopy obscuration. While Light Detection and Ranging (LIDAR) systems provide a way to penetrate through the canopy, the limited spatial samples on the ground greatly reduce the efficacy of detection and classification algorithms. Hyperspectral sensors provide the ability to perform subpixel detection, but due to their passive nature, very few photons from below the canopy reach the sensor. Ideally, a hyperspectral LIDAR system would provide the best of both worlds. The LIDAR system would provide penetration through the foliage while the spectral system would provide improved classification ability. Currently, such sensor systems are in their infancy, but advances are being made in technology that would allow for their widespread development. As a proof of concept, we develop a model to simulate the results and show the benefit they bring to FOPEN applications. We start with basic shapes and randomly remove points to create a situation similar to FOPEN. Different spectral and spatial combinations are evaluated to determine what gains in performance are possible by adding spectral information to LIDAR systems.