RADIOMETRIC CALIBRATION OF AIRBORNE LIDAR INTENSITY DATA FOR LAND COVER CLASSIFICATION

The rapid development of the airborne LiDAR systems paves the way for the use of the LiDAR technology in different bathymetric and topographic applications. LiDAR has been used effectively for digital terrain/surface modelling by measuring the range from the sensor to the earth surface. Information can be extracted on the geometry of the scanned features (e.g. buildings, roads) or surfaces elevation from the 3D point cloud. Yet, few studies explored the use of the LiDAR intensity data with the ranging data for land cover classification. LiDAR intensity is recorded as the amount of energy backscattered from objects. Generally, it requires certain radiometric calibration scheme to calibrate and correct the reflected intensity data of the land cover features to its physical spectral reflectance. Approximate approaches are proposed for the radiometric calibration of the LiDAR intensity including noise filtering, empirical modelling and cross-sensor calibration. Nevertheless, none of them demonstrates a viable solution for fast and accurate calibration. In this regard, some researches proposed to model the radar equation for radiometric calibration of the LiDAR intensity data. This study demonstrates how to radiometrically correct the LiDAR intensity data based on the range, the incidence angle of the laser beam, the laser footprint, the slope of the target, the ground object reflectance, and the atmospheric attenuation. The long term goal of the research is to devise a fast and accurate radiometric calibration scheme for the airborne LiDAR intensity data (both multi-echo and full-waveform) in order to maximize the use of the intensity data in large scale national land cover mapping. The objective of the calibration is to enhance the class separability amongst different land cover types before classifying the LiDAR data intensity. The range and incidence angle can be retrieved from the raw data of laser point cloud. The footprint area and the slope of the target are dependent on the topography of the study area. Homogeneous ground objects are identified in the study area in order to link them with the corresponding spectrum reflectance from the U.S. Geological Survey digital spectral library. The atmospheric attenuation can be derived by a sophisticated meteorological model obtained from Government weather station. All the parameters of the radar equation are applied to correct and calibrate the received energy for each laser pulse. After the radiometric calibration, a set of feature vectors (including the elevation data) is constructed derived from airborne LiDAR data. Both pixel-based and object-based classifiers can be applied on the calibrated data for land cover classification.