Multispectral LiDAR data for land cover classification of urban areas

Ahmed Abdelrahman,Ahmed El-Rabbany,Salem Morsy

doi:10.32920/14638104

Abstract

Airborne Light Detection And Ranging (LiDAR) systems usually operate at a monochromatic wavelength measuring the range and the strength of the reflected energy (intensity) from objects. Recently, multispectral LiDAR sensors, which acquire data at different wavelengths, have emerged. This allows for recording of a diversity of spectral reflectance from objects. In this context, we aim to investigate the use of multispectral LiDAR data in land cover classification using two different techniques. The first is image-based classification, where intensity and height images are created from LiDAR points and then a maximum likelihood lassifier is applied. The second is point-based classification, where ground filtering and Normalized Difference Vegetation Indices (NDVIs) computation are conducted. A dataset of an urban area located in Oshawa, Ontario, Canada, is classified into four classes: buildings, trees, roads and grass. An overall accuracy of up to 89.9% and 92.7% is achieved from image classification and 3D point classification, respectively. A radiometric correction model is also applied to the intensity data in order to remove the attenuation due to the system distortion and terrain height variation. The classification process is then repeated, and the results demonstrate that there are no significant improvements achieved in the overall accuracy. Keywords: multispectral LiDAR; land cover; ground filtering; NDVI; radiometric correction

Highlights

With the evolution of airborne Light Detection And Ranging (LiDAR) technology, numerous studies have been conducted on the use of airborne LiDAR height and intensity data for land cover classification [1–4]
We aim to present the capability of using multispectral airborne LiDAR data for land cover classification
Previous studies achieved overall classification accuracies from 85–89.5% for the same four land cover classes. They used multispectral aerial/satellite imagery combined with normalized Digital Surface Model (nDSM) derived from LiDAR data [5,6] or combined with LiDAR height and intensity data [7–9], while the presented work in this research used the LiDAR data only

Summary

Introduction

With the evolution of airborne LiDAR technology, numerous studies have been conducted on the use of airborne LiDAR height and intensity data for land cover classification [1–4]. The normalized Digital Surface Model (nDSM) with multispectral aerial/satellite imagery [5,6]. Other investigations have combined multispectral aerial/satellite imagery with LiDAR height and intensity data [7–9]. Since most of the previous studies converted either LiDAR intensity or height data into 2D images, typical LiDAR images such as intensity [2–4,7,8], multiple returns [2,7], DSM, the Digital Terrain Model (DTM) [2,4] and nDSM [5,6,8] were created. When the LiDAR intensity data were combined with multispectral aerial/satellite imagery, the NDVI was used [5,6,8]. Traditional supervised pixel-based classification techniques such as maximum likelihood [5,9], rule-based classification [2,3,8] and the Gaussian mixture model [7] were applied. Other studies accounted for the spatial coherence of different objects to avoid the noises in the pixel-based classification results by using object-orientated classification techniques [4,6]

Objectives

Methods

Results

Discussion

Conclusion