LiDAR Ground Research Articles

Modeling Merchantable Wood Volume Using Airborne LiDAR Metrics and Historical Forest Inventory Plots at a Provincial Scale

So far, large-scale projects aiming to map forest attributes using aerial LiDAR data have been developed using ground sample plots acquired synchronously with LiDAR. No large projects have been developed using aerial LiDAR acquired independent of ground sample plot datasets. The goal of this study was to develop and validate large-scale parametric merchantable wood volume estimation models using existing historical ground sample plots. The models can be applied to large LiDAR datasets to map merchantable wood volume as a 10 × 10 m raster. The study demonstrated that a relative density index (RDI) based on a self-thinning equation and dominant height were suitable variables that can be calculated both for ground sample plots and LiDAR datasets. The resulting volume raster showed sound accuracy rates when compared to validation zones: R², 82.25%; RMSE, 13.7 m3/ha; and bias, −4.09 m3/ha. The results show that ground sample plot datasets acquired synchronously with LiDAR can be used to calculate the RDI and dominant height. These variables can consequently be used to map forest attributes over a large area with a high level of accuracy, thus not requiring the implementation of new costly sample plots.

A Robust Gaussian Process-Based LiDAR Ground Segmentation Algorithm for Autonomous Driving

Robust and precise vehicle detection is the prerequisite for decision-making and motion planning in autonomous driving. Vehicle detection algorithms follow three steps: ground segmentation, obstacle clustering and bounding box fitting. The ground segmentation result directly affects the input of the subsequent obstacle clustering algorithms. Aiming at the problems of over-segmentation and under-segmentation in traditional ground segmentation algorithms, a ground segmentation algorithm based on Gaussian process is proposed in this paper. To ensure accurate search of real ground candidate points as training data for Gaussian process, the proposed algorithm introduces the height and slope criteria, which is more reasonable than the use of fixed height threshold for searching. After that, a sparse covariance function is introduced as the kernel function for calculation in Gaussian process. This function is more suitable for ground segmentation situation the radial basis function (RBF). The proposed algorithm is tested on our autonomous driving experimental platform and the public autonomous driving dataset KITTI, compared with the most used RANSAC algorithm and ray ground filter algorithm. Experiment results show that the proposed algorithm can avoid obvious over-segmentation and under-segmentation. In addition, compared with the RBF, the introduction of the sparse covariance function also reduces the computation time by 37.26%.

LiDAR Filtering in 3D Object Detection Based on Improved RANSAC

At present, the LiDAR ground filtering technology is very mature. There are fewer applications in 3D-object detection due to the limitations of filtering accuracy and efficiency. If the ground can be removed quickly and accurately, the 3D-object detection algorithm can detect objects more accurately and quickly. In order to meet the application requirements of 3D-object detection, inspired by Universal-RANSAC, we analyze the detailed steps of RANSAC and propose a precise and efficient RANSAC-based ground filtering method. The principle of GroupSAC is analyzed, and the sampled points are grouped by attributes to make it easier to sample the correct point. Based on this principle, we devise a method for limiting sampled points that is applicable to point clouds. We describe preemptive RANSAC in detail. Its breadth-first strategy is adopted to obtain the optimal plane without complex iterations. We use the International Society for Photogrammetry and Remote Sensing (ISPRS) datasets and the KITTI dataset for testing. Experiments show that our method has higher filtering accuracy and efficiency compared with the currently widely used methods. We explore the application of ground filtering methods in 3D-object detection, and the experimental results show that our method can improve the object detection accuracy without affecting the efficiency.

Vertical aerosol data assimilation technology and application based on satellite and ground lidar: A review and outlook

Observations and numerical models are mainly used to investigate the spatiotemporal distribution and vertical structure characteristics of aerosols to understand aerosol pollution and its effects. However, the limitations of observations and the uncertainties of numerical models bias aerosol calculations and predictions. Data assimilation combines observations and numerical models to improve the accuracy of the initial, analytical fields of models and promote the development of atmospheric aerosol pollution research. Numerous studies have been conducted to integrate multi-source data, such as aerosol optical depth and aerosol extinction coefficient profile, into various chemical transport models using various data assimilation algorithms and have achieved good assimilation results. The definition of data assimilation and the main algorithms will be briefly presented, and the progress of aerosol assimilation according to two types of aerosol data, namely, aerosol optical depth and extinction coefficient, will be presented. The application of vertical aerosol data assimilation, as well as the future trends and challenges of aerosol data assimilation, will be further analysed.

Applying a Robust Empirical Method for Comparing Repeated LiDAR Data with Different Point Density

A key aspect of vegetation monitoring from LiDAR is concerned with the use of comparable data acquired from multitemporal surveys and from different sensors. Accurate digital elevation models (DEMs) to derive vegetation products, are required to make comparisons among repeated LiDAR data. Here, we aimed to apply an improved empirical method based on DEMs of difference, that adjust the ground elevation of a low-density LiDAR dataset to that of a high-density LiDAR one for ensuring credible vegetation changes. The study areas are a collection of six sites over the Sierra de Gredos in Central Spain. The methodology consisted of producing “the best DEM of difference” between low- and high-density LiDAR data (using the classification filter, the interpolation method and the spatial resolution with the lowest vertical error) to generate a local “pseudo-geoid” (i.e., continuous surfaces of elevation differences) that was used to correct raw low-density LiDAR ground points. The vertical error of DEMs was estimated by the 50th percentile (P50), the normalized median absolute deviation (NMAD) and the root mean square error (RMSE) of elevation differences. In addition, we analyzed the effects of site-properties (elevation, slope, vegetation height and distance to the nearest geoid point) on DEMs accuracy. Finally, we assessed if vegetation height changes were related to the ground elevation differences between low- and high-density LiDAR datasets. Before correction and aggregating by sites, the vertical error of DEMs ranged from 0.02 to −2.09 m (P50), from 0.39 to 0.85 m (NMDA) and from 0.54 to 2.5 m (RMSE). The segmented-based filter algorithm (CSF) showed the highest error, but there were not significant differences among interpolation methods or spatial resolutions. After correction and aggregating by sites, the vertical error of DEMs dropped significantly: from −0.004 to −0.016 m (P50), from 0.10 to 0.06 m (NMDA) and from 0.28 to 0.46 m (RMSE); and the CSF filter algorithm continued showing the greatest vertical error. The terrain slope and the distance to the nearest geoid point were the most important variables for explaining vertical accuracy. After corrections, changes in vegetation height were decoupled from vertical errors of DEMs. This work showed that using continuous surfaces with the lowest elevation differences (i.e., the best DEM of difference) the raw elevation of low-density LiDAR was better adjusted to that of a benchmark for being adapted to site-specific conditions. This method improved the vertical accuracy of low-density LiDAR elevation data, minimizing the random nature of vertical errors and decoupling vegetation changes from those errors.

Plant functional remote sensing and smart farming applications

Plants have the distinctive 3D spatial structure that varies among organs, species and communities, and the spatial structure changes as they interact with their environments. The functions linked to fundamental biological activities such as transpiration, photosynthesis, and growth are also affected by the spatial structure and the environment. In order to promote smart farming using information and communication technology (ICT), it is necessary to measure and utilize information at the cell-organ of plants to the individual and the community levels and the environments in two or even three dimensions. Therefore, this paper introduced the outline of remote sensing of plant functioning and examples of the 3D remote sensing from relatively short distances using drones and ground Lidar. The quality control of rice in the paddy field and chlorophyll fluorescence imaging for photosynthetic diagnosis were also introduced. In addition, a field smart farm and a smart greenhouse, which heavily utilize ICT, built at Takasaki University of Health and Welfare in Gunma, Japan, were also introduced. Keywords: remote sensing, smart farming, smart greenhouse, IoT, ICT, image analysis DOI: 10.25165/j.ijabe.20221504.7375 Citation: Omasa K, Ono E, Ishigami Y, Shimizu Y, Araki Y. Plant functional remote sensing and smart farming applications. Int J Agric & Biol Eng, 2022; 15(4): 1–6.

Aerosol Physical‐Optical Properties and PBLH Under Different Air Pollution Levels From Ground Lidar and Satellite Observations Over Shouxian Area, China

AbstractThe physical and optical properties of aerosols and the change of planetary boundary layer height (PBLH) are of great scientific significance for the study of atmospheric environment and air quality. This paper intends to present integrated observations of aerosol physical‐optical characteristics and PBLH properties under different air pollution levels in Shouxian (National Climatology Observatory (32°26’ N, 116°47’ E)), China. The combined measurements of ground‐based Lidar and Cloud‐Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) are analyzed from December 2016 to November 2017 in this study. Results show that the extinction coefficients decrease significantly with height below 2 km. The extinction coefficients below 1 km, the aerosol optical depth (AOD), and the PBLH increase with the aggravation of pollution. The daily average aerosol extinction coefficient near the surface is about 0.78 km−1, the AOD is 0.87 ± 0.33 m (mean ± std, the same below), and the PBLH is 827 ± 132 m in polluted conditions (PM2.5 > 115 μg m−3). With the increase in pollution and the decrease in PBLH, more aerosols are accumulated at a lower height. The daily average ratios of AOD below PBLH (AODPBLH) to total AOD (AODtot) are about 64% in polluted conditions. According to the CALIPSO‐observed aerosol vertical profile, aerosols are identified for about 20.54% of the vertical space below 6 km altitude under polluted conditions. The spherical fine particles with lower particulate depolarization ratio (PDR, 79% of the PDR is <0.2), lower color ratio (CR, 64% of the CR is <0.75), and stronger scattering ability are dominant in 0∼2 km.

High-Resolution Satellite Stereo Matching by Object-Based Semiglobal Matching and Iterative Guided Edge-Preserving Filter

This letter presents an object-based stereo matching and content-based guided disparity map refinement for high-resolution satellite stereo images. The proposed method uses the superpixels for object-based cost filtering and then object-based semiglobal matching. In the proposed object-based stereo matching strategy, two improvements including homogeneity weight for cost filtering, and weighted cost aggregation by image objects are considered. After cost aggregation, the generated disparity map is computed using the winner takes all and then refined with a new iterative guided edge-preserving filter. The proposed method with Census cost function has been implemented on high-resolution satellite stereo images and then is compared with the LiDAR ground truth. Moreover, the proposed method is compared with five state-of-the-art stereo matching methods. The experimental results on along-track satellite stereo images from Pleiades and IKONOS images and cross-track multidate images from WV-III demonstrate that the proposed method significantly improves the result of stereo matching.

Semantic segmentation on Swiss3DCities: A benchmark study on aerial photogrammetric 3D pointcloud dataset

We introduce a new outdoor urban 3D pointcloud dataset, covering a total area of 2.7km2, sampled from three Swiss cities with different characteristics. The dataset is manually annotated for semantic segmentation with per-point labels, and is built using photogrammetry from images acquired by multirotors equipped with high-resolution cameras. In contrast to datasets acquired with ground LiDAR sensors, the resulting point clouds are uniformly dense and complete, and are useful to disparate applications, including autonomous driving, gaming and smart city planning. As a benchmark, we report quantitative results of PointNet++, an established point-based deep 3D semantic segmentation model; on this model, we additionally study the impact of using different cities for model generalization.

Vertical structure heterogeneity in broadleaf forests: Effects on light interception and canopy photosynthesis

Foliage clumping refers to the aggregation of leaves around branches and tree crowns. It has been shown to influence considerably light interception in forests and is increasingly being used in terrestrial biosphere models (TBMs) to estimate canopy photosynthesis, represented by a single value encompassing all spatial scales (shoot, branch, crown and plot): the clumping factor (Ω). Several studies have pointed to possible vertical variations in foliage clumping, and a recent study confirmed very low clumping factor values (high leaf aggregation) at the top of deciduous broadleaf forests. However complete profiles of clumping factors have never been described in tall forest canopies. As TBMs are currently moving towards multilayer schemes for the computation of photosynthesis, we investigated whether vertical profiles of foliage clumping should be considered within those schemes. We first used ground lidar combined with hemispherical photos to characterise branch level clumping factor (Ωv) at different heights in a deciduous broadleaf forest. We then applied the branch level clumping factor profile to four forests plots where complete 3D structure derived from ground lidar was available to produce vertical profiles (30 cm resolution) for foliage clumping (Ωl) and leaf area index (LAI). We ran radiative transfer simulations using different scenarios for representing vertical structure to assess its effect on canopy photosynthesis (without considering energy balance, temperature, kinetics or water balance). We found that when considering branch level clumping, the plot level clumping factor values were considerably lower (leaves more clumped together) than the average values commonly used for broadleaf forests. Modeling light fluxes using this lower value increased canopy photosynthesis because of a combination of greater light penetration to lower layers and non-linearity in the light-response curve. Considering the vertical profiles of foliage clumping further increased canopy photosynthesis through a greater contribution from shaded leaves in the lower and upper canopy levels. Further, we find that there is an optimal range of value for clumping factor which maximises photosynthesis, and the four sites surveyed (LAIs above 4) display clumping factors values close to this optimum. Solely on the basis of the effect of radiative transfer, we did not find notable differences in canopy photosynthesis when the LAI profile was considered. We find that it may be of interest to consider foliage clumping vertical profiles within TBM multilayer schemes, and we present a generic profile equation for use in dense deciduous forests.

The Effect of LiDAR Sampling Density on DTM Accuracy for Areas with Heavy Forest Cover

Laser scanning via LiDAR is a powerful technique for collecting data necessary for Digital Terrain Model (DTM) generation, even in densely forested areas. LiDAR observations located at the ground level can be separated from the initial point cloud and used as input for the generation of a Digital Terrain Model (DTM) via interpolation. This paper proposes a quantitative analysis of the accuracy of DTMs (and derived slope maps) obtained from LiDAR data and is focused on conditions common to most forestry activities (rough, steep terrain with forest cover). Three interpolation algorithms were tested: Inverse Distance Weighted (IDW), Natural Neighbour (NN) and Thin-Plate Spline (TPS). Research was mainly focused on the issue of point data density. To analyze its impact on the quality of ground surface modelling, the density of the filtered data set was artificially lowered (from 0.89 to 0.09 points/m2) by randomly removing point observations in 10% increments. This provides a comprehensive method of evaluating the impact of LiDAR ground point density on DTM accuracy. While the reduction of point density leads to a less accurate DTM in all cases (as expected), the exact pattern varies by algorithm. The accuracy of the LiDAR-derived DTMs is relatively good even when LiDAR sampling density is reduced to 0.40–0.50 points/m2 (50–60 % of the initial point density), as long as a suitable interpolation algorithm is used (as IDW proved to be less resilient to density reductions below approximately 0.60 points/m2). In the case of slope estimation, the pattern is relatively similar, except the difference in accuracy between IDW and the other two algorithms is even more pronounced than in the case of DTM accuracy. Based on this research, we conclude that LiDAR is an adequate method for collecting morphological data necessary for modelling the ground surface, even when the sampling density is significantly reduced.

基于点云簇组合特征的激光雷达地面分割方法

基于点云簇组合特征的激光雷达地面分割方法

Fusion of ground penetrating radar and laser scanning for infrastructure mapping

Abstract Mobile mapping vehicles, equipped with cameras, laser scanners (in this paper referred to as light detection and ranging, LiDAR), and positioning systems are limited to acquiring surface data. However, in this paper, a method to fuse both LiDAR and 3D ground penetrating radar (GPR) data into consistent georeferenced point clouds is presented, allowing imaging both the surface and subsurface. Objects such as pipes, cables, and wall structures are made visible as point clouds by thresholding the GPR signal’s Hilbert envelope. The results are verified with existing utility maps. Varying soil conditions, clutter, and noise complicate a fully automatized approach. Topographic correction of the GPR data, by using the LiDAR data, ensures a consistent ground height. Moreover, this work shows that the LiDAR point cloud, as a reference, increases the interpretability of GPR data and allows measuring distances between above ground and subsurface structures.

Determination of atmospheric pollution of megapolis by space data

At present, air pollution is one of the most serious environmental problems. To monitor the air condition, sampling methods are used at stationary and automatic stations, lidar ground measurements and satellite monitoring. In Belarus, the atmosphere is monitored at 66 stationary stations of the National Environmental Monitoring System of the Republic of Belarus. The main disadvantage of the sampling method at stations is its localization and inoperability. The method implemented in the work overcomes these shortcomings. Estimates of the atmospheric pollution of megacities in units of maximum permissible concentrations (MPC) are carried out using the space reflection spectrum and the image in red region of the visible spectrum. Using data from the EO-1, Landsat, Sentinel-2 satellites, as well as calculations, the integral pollution levels (in MPC) over the city of Minsk were obtained in 2015 and 2016. Estimated pollution values averaged over the quarter were compared with data from the annual of the status of atmospheric air of the National Environmental Monitoring System of the Republic of Belarus. The method showed good agreement between the data. However, the conditions of application and the accuracy of this method require further research and evaluation, the advantages and disadvantages of the method are analyzed in the work, and ways to overcome the identified shortcomings are outlined.

The effect of leaf-on and leaf-off forest canopy conditions on LiDAR derived estimations of forest structural diversity

Forest structural diversity metrics describing diversity in tree size and crown shape within forest stands can be used as indicators of biodiversity. These diversity metrics can be generated using airborne laser scanning (LiDAR) data to provide a rapid and cost effective alternative to ground-based inspection. Measures of tree height derived from LiDAR can be significantly affected by the canopy conditions at the time of data collection, in particular whether the canopy is under leaf-on or leaf-off conditions, but there have been no studies of the effects on structural diversity metrics. The aim of this research is to assess whether leaf-on/leaf-off changes in canopy conditions during LiDAR data collection affect the accuracy of calculated forest structural diversity metrics. We undertook a quantitative analysis of LiDAR ground detection and return height, and return height diversity from two airborne laser scanning surveys collected under leaf-on and leaf-off conditions to assess initial dataset differences. LiDAR data were then regressed against field-derived tree size diversity measurements using diversity metrics from each LiDAR dataset in isolation and, where appropriate, a mixture of the two. Models utilising leaf-off LiDAR diversity variables described DBH diversity, crown length diversity and crown width diversity more successfully than leaf-on (leaf-on models resulted in R² values of 0.66, 0.38 and 0.16, respectively, and leaf-off models 0.67, 0.37 and 0.23, respectively). When LiDAR datasets were combined into one model to describe tree height diversity and DBH diversity the models described 75% and 69% of the variance (R² of 0.75 for tree height diversity and 0.69 for DBH diversity). The results suggest that tree height diversity models derived from airborne LiDAR, collected (and where appropriate combined) under any seasonal conditions, can be used to differentiate between simple single and diverse multiple storey forest structure with confidence.

Monitoring of Morphological Changes in an Estuary based on Time-Series Analysis of Sediment Discharge Using a Hydrophone

Jun, K.W.; Kin, S.D.; Lee, H.J., and Choi, J.H., 2020. Monitoring of morphological changes in an estuary based on time-series analysis of sediment discharge using a hydrophone. In: Malvárez, G. and Navas, F. (eds.), Global Coastal Issues of 2020. Journal of Coastal Research, Special Issue No. 95, pp. 516–521. Coconut Creek (Florida), ISSN 0749-0208.The purpose of this study is to predict the littoral transport rate and morphological changes of estuaries by measuring the discharge of sediments from rivers into the sea in a time series using a hydrophone. To identify the problem of longshore sediments, an understanding of waves and currents is necessary. However, the runoff characteristics of sediments in river estuary that serve as the main source of sediments should be grasped first. It is important to predict the discharge of sediments from rivers into the sea and understand the coastal processes involved in the formation of sandbars in the estuary and the transport of sediments to coastal waters. The selected study area was the Gagok River located in the East Coast of Korea, where periodic estuary occlusion occurs due to the behavior of rivers and waves. This study investigated the morphological changes in the estuary according to the changes in sediment discharge. The sediment discharge was measured using a hydrophone capable of measuring the continuous sediment transport from the acoustic information based on the movement of the sediments. Based on this method, the hydrophone was applied to the field, and measurements of major rainfall events were taken to investigate the influence of changes in sediment discharge on the morphological changes in the estuary. In addition, to verify the results, the morphological changes in the estuary were linked to the ground LiDAR data measured over several years. The results of the survey confirmed that the occlusion phenomenon of the Gagok River estuary occurred according to the seasonal changes. Except for the flood season, there was a small amount of runoff through the river. As a result, a sand barrier developed in the estuary due to the influence of the sediments transported by the waves from the open sea.

Influence of digital elevation models on the simulation of rainfall-induced landslides in the hillslopes of Guwahati, India

Topographic input parameters such as slope, curvature, and drainage area are derived from a Digital Elevation Model (DEM) representing the spatial elevation, and are widely used as important sources of geospatial information. Based on the chosen DEM and its derived paramters, physically based GIS models, such as TRIGRS, can compute the transient degradation of the hillslope stability due to rainfall infiltration to identify the landslide occurrences in the considered region. Hence, it is inadvertent that the accuracy of the DEM will significantly affect the outcome of the TRIGRS simulations. Obtaining a high-resolution DEM (using LiDAR, dGPS or other such ground based advanced surveying methods) for a large area is an expensive affair and unavailable in most cases. Hence, under such circumstances, DEMs available in public domain vis., CartoDEM, ALOS-AW3D30-DEM, SRTM-DEM, and ASTER-GDEM, serves to be the sole option. Each of these DEMs are obtained from varying sources using different data collection and processing techniques, thereby leading to DEMs of varying resolutions and different accuracies. Although there is ample literature on the effects of the resolution of specific DEMs, the influence of the varying accuracies of DEMs chosen for TRIGRS simulation has not been adequately reported. Considering the landslide scenario of Guwahati city as the study region, the effect of dissimilarity in DEMs on the prediction of the rainfall-induced landslides in the hillslopes of the study area is reported in this paper. Three rainfall events triggering landslides were considered as input into TRIGRS, and the simulation outcome of the different DEMs were compared with reported landslide locations. Significantly different outputs were obtained for the DEMs highlighting the importance of the analysis. Based on two different evaluation methodologies, i.e. the ROC and LRclass, landslide occurrences identified by ALOS-3D DEM provided the best agreements with the reported landslides, followed by those identified by Carto-DEMs.

A Point-Based Fully Convolutional Neural Network for Airborne LiDAR Ground Point Filtering in Forested Environments

Airborne laser scanning (ALS) data is one of the most commonly used data for terrain products generation. Filtering ground points is a prerequisite step for ALS data processing. Traditional filtering methods mainly use handcrafted features or predefined classification rules with preprocessing/post-processing operations to filter ground points iteratively, which is empirical and cumbersome. Deep learning provides a new approach to solve classification and segmentation problems because of its ability to self-learn features, which has been favored in many fields, particularly remote sensing. In this article, we proposed a point-based fully convolutional neural network (PFCN) which directly consumed points with only geometric information and extracted both point-wise and tile-wise features to classify each point. The network was trained with 37449157 points from 14 sites and evaluated on 6 sites in various forested environments. Additionally, the method was compared with five widely used filtering methods and one of the best point-based deep learning methods (PointNet++). Results showed that the PFCN achieved the best results in terms of mean omission error (T1 = 1.10%), total error (Te = 1.73%), and Kappa coefficient (93.88%), but ranked second for the root mean square error of the digital Terrain model caused by the worst commission error. Additionally, our method was on par with or even better than PointNet++ in accuracy. Moreover, the method consumes one-third of the computational resource and one-seventh of the training time. We believe that PFCN is a simple and flexible method that can be widely applied for ground point filtering.

Building 3D Realistic Modeling Based on Air-ground Multi-source Data Fusion

Three-dimensional point cloud data is a new form of three-dimensional collection, which not only contains the geometric topology information of the object, but also has high simplicity and flexibility. In this paper, the air-ground multi-source data fusion technology is used to study the fine reconstruction of the 3D scene: based on the 3D laser scanning laser point cloud, the 3D spatial information of the ground visible objects is obtained, and the orthophoto obtained by the drone aerial photography is assisted, Obtain the three-dimensional space information of the top of the ground feature, and the ground three-dimensional laser scanner can quickly obtain the three-dimensional surface information of the building facade, ground, and trees. Due to the complex structure of the building and the occlusion of spatial objects, sub-station scanning is required when acquiring point cloud data. This article uses the Sino-German Energy Conservation Center Building of Shenyang Jianzhu University as the research area, using drone tilt photography technology and ground lidar technology to integrate. During the experiment, the field industry adopted the UAV image acquisition strategy of “automatic shooting of regular routes, supplemented by manual shooting of areas of interest”; in the field industry, the method of “manual coarse registration and ICP algorithm fine registration” The example results show that the ground 3D laser point cloud air-ground image fusion 3D modeling effect proposed in this paper is better and the quality is greatly improved, which makes up for the ground 3D laser scanning. In point cloud modeling, a large number of holes are insufficient due to occlusion and missing top information.

Method for 3-D Scene Reconstruction Using Fused LiDAR and Imagery From a Texel Camera

Reconstructing a 3-D scene from aerial sensor data creating a textured digital surface model (TDSM), consisting of a LiDAR point cloud and an overlaid image, is valuable in many applications including agriculture, military, surveying, and natural disaster response. When collecting LiDAR from an aircraft, the navigation system accuracy must exceed the LiDAR accuracy to properly reference returns in 3-D space. Precision navigation systems can be expensive and often require full-scale aircraft to house such systems. Synchronizing the LiDAR sensor and a camera, using a texel camera calibration, provides additional information that reduces the need for precision navigation equipment. This paper describes a bundle adjustment technique for aerial texel images that allows for relatively low-accuracy navigation systems to be used with low-cost LiDAR and camera data to form higher fidelity terrain models. The bundle adjustment objective function utilizes matching image points, measured LiDAR distances, and the texel camera calibration and does not require overlapping LiDAR scans or ground control points. The utility of this method is proven using a simulated texel camera and unmanned aerial system (UAS) flight data created from aerial photographs and elevation data. A small UAS is chosen as the target vehicle due to its relatively inexpensive hardware and operating costs, illustrating the power of this method in accurately referencing the LiDAR and camera data. In the 3-D reconstruction, the 1- $\sigma $ accuracy between LiDAR measurements across the scene is on the order of the digital camera pixel size.