Estimation of Optimal Parameter for Range Normalization of Multispectral Airborne Lidar Intensity Data

Abstract

Abstract. Range normalization is a common data pre-process that aims to improve the radiometric quality of airborne LiDAR data. This radiometric treatment considers the rate of energy attenuation sustained by the laser pulse as it travels through a medium back and forth from the LiDAR system to the surveyed object. As a result, the range normalized intensity is proportional to the range to the power of a factor a. Existing literature recommended different a values on different land cover types, which are commonly adopted in forestry studies. Nevertheless, there is a lack of study evaluating the range normalization on multispectral airborne LiDAR intensity data. In this paper, we propose an overlap-driven approach that is able to estimate the optimal a value by pairing up the closest data points of two overlapping LiDAR data strips, and subsequently estimating the range normalization parameter a based on a least-squares adjustment. We implemented the proposed method on a set of multispectral airborne LiDAR data collected by a Optech Titan, and assessed the coefficient of variation of four land cover types before and after applying the proposed range normalization. The results showed that the proposed method was able to estimate the optimal a value, yielding the lowest cv, as verified by a cross validation approach. Nevertheless, the estimated a value is never identical for the four land cover classes and the three laser wavelengths. Therefore, it is not recommended to label a specific a value for the range normalization of airborne LiDAR intensity data within a specific land cover type. Instead, the range normalization parameter is deemed to be data-driven and should be estimated for each LiDAR dataset and study area.

Proximal and remote sensing in plant phenomics: 20 years of progress, challenges, and perspectives

“…2D image–based phenotyping (e.g., with RGB images or multi-/hyperspectral images) usually involves image registration ( Tondewad and Dale, 2020 ), classification ( Cheng et al., 2020 ), segmentation ( Hossain and Chen, 2019 ), and trait extraction ( Jiang et al., 2020 ). 3D data, such as LiDAR or image-reconstructed point clouds, typically undergo a processing pipeline of registration ( Cheng et al., 2018 ), denoising ( Hu et al., 2021 ), sampling ( Bergman et al., 2020 ), filtering ( Jin et al., 2020a ), normalization ( Kwan and Yan, 2020 ), classification/segmentation ( Mao and Hou, 2019 ), and trait extraction ( Jin et al., 2021b ). These phenotyping methods enable the extraction of structural, physiological, and performance-related traits.…”