Light Detection And Ranging Point Research Articles

Growth Height Determination of Tree Walls for Precise Monitoring in Apple Fruit Production Using UAV Photogrammetry

In apple cultivation, spatial information about phenotypic characteristics of tree walls would be beneficial for precise orchard management. Unmanned aerial vehicles (UAVs) can collect 3D structural information of ground surface objects at high resolution in a cost-effective and versatile way by using photogrammetry. The aim of this study is to delineate tree wall height information in an apple orchard applying a low-altitude flight pattern specifically designed for UAVs. This flight pattern implies small distances between the camera sensor and the tree walls when the camera is positioned in an oblique view toward the trees. In this way, it is assured that the depicted tree crown wall area will be largely covered with a larger ground sampling distance than that recorded from a nadir perspective, especially regarding the lower crown sections. Overlapping oblique view images were used to estimate 3D point cloud models by applying structure-from-motion (SfM) methods to calculate tree wall heights from them. The resulting height models were compared with ground-based light detection and ranging (LiDAR) data as reference. It was shown that the tree wall profiles from the UAV point clouds were strongly correlated with the LiDAR point clouds of two years (2018: R2 = 0.83; 2019: R2 = 0.88). However, underestimation of tree wall heights was detected with mean deviations of −0.11 m and −0.18 m for 2018 and 2019, respectively. This is attributed to the weaknesses of the UAV point clouds in resolving the very fine shoots of apple trees. Therefore, the shown approach is suitable for precise orchard management, but it underestimated vertical tree wall expanses, and widened tree gaps need to be accounted for.

Accelerated LiDAR data processing algorithm for self‐driving cars on the heterogeneous computing platform

In recent years, light detection and ranging (LiDAR) has been widely used in the field of self-driving cars, and the LiDAR data processing algorithm is the core algorithm used for environment perception in self-driving cars. At the same time, the real-time performance of the LiDAR data processing algorithm is highly demanding in self-driving cars. The LiDAR point cloud is characterised by its high density and uneven distribution, which poses a severe challenge in the implementation and optimisation of data processing algorithms. In view of the distribution characteristics of LiDAR data and the characteristics of the data processing algorithm, this study completes the implementation and optimisation of the LiDAR data processing algorithm on an NVIDIA Tegra X2 computing platform and greatly improves the real-time performance of LiDAR data processing algorithms. The experimental results show that compared with an Intel® Core™ i7 industrial personal computer, the optimised algorithm improves feature extraction by nearly 4.5 times, obstacle clustering by nearly 3.5 times, and the performance of the whole algorithm by 2.3 times.

CNN-based 3D object classification using Hough space of LiDAR point clouds

With the wide application of Light Detection and Ranging (LiDAR) in the collection of high-precision environmental point cloud information, three-dimensional (3D) object classification from point clouds has become an important research topic. However, the characteristics of LiDAR point clouds, such as unstructured distribution, disordered arrangement, and large amounts of data, typically result in high computational complexity and make it very difficult to classify 3D objects. Thus, this paper proposes a Convolutional Neural Network (CNN)-based 3D object classification method using the Hough space of LiDAR point clouds to overcome these problems. First, object point clouds are transformed into Hough space using a Hough transform algorithm, and then the Hough space is rasterized into a series of uniformly sized grids. The accumulator count in each grid is then computed and input to a CNN model to classify 3D objects. In addition, a semi-automatic 3D object labeling tool is developed to build a LiDAR point clouds object labeling library for four types of objects (wall, bush, pedestrian, and tree). After initializing the CNN model, we apply a dataset from the above object labeling library to train the neural network model offline through a large number of iterations. Experimental results demonstrate that the proposed method achieves object classification accuracy of up to 93.3% on average.

Roof Plane Segmentation from Airborne LiDAR Data Using Hierarchical Clustering and Boundary Relabeling

The roof plane segmentation is one of the key issues for constructing accurate three-dimensional building models from airborne light detection and ranging (LiDAR) data. Region growing is one of the most widely used methods to detect roof planes. It first selects one point or region as a seed, and then iteratively expands to neighboring points. However, region growing has two problems. The first problem is that it is hard to select the robust seed points. The other problem is that it is difficult to detect the accurate boundaries between two roof planes. In this paper, to solve these two problems, we propose a novel approach to segment the roof planes from airborne LiDAR point clouds using hierarchical clustering and boundary relabeling. For the first problem, we first extract the initial set of robust planar patches via an octree-based method, and then apply the hierarchical clustering method to iteratively merge the adjacent planar patches belonging to the same plane until the merging cost exceeds a predefined threshold. These merged planar patches are regarded as the robust seed patches for the next region growing. The coarse roof planes are generated by adding the non-planar points into the seed patches in sequence using region growing. However, the boundaries of coarse roof planes may be inaccurate. To solve this problem, namely, the second problem, we refine the boundaries between adjacent coarse planes by relabeling the boundary points. At last, we can effectively extract high-quality roof planes with smooth and accurate boundaries from airborne LiDAR data. We conducted our experiments on two datasets captured from Vaihingen and Wuhan using Leica ALS50 and Trimble Harrier 68i, respectively. The experimental results show that our proposed approach outperforms several representative approaches in both visual quality and quantitative metrics.

A Fast Spatial Clustering Method for Sparse LiDAR Point Clouds Using GPU Programming.

Fast and accurate obstacle detection is essential for accurate perception of mobile vehicles’ environment. Because point clouds sensed by light detection and ranging (LiDAR) sensors are sparse and unstructured, traditional obstacle clustering on raw point clouds are inaccurate and time consuming. Thus, to achieve fast obstacle clustering in an unknown terrain, this paper proposes an elevation-reference connected component labeling (ER-CCL) algorithm using graphic processing unit (GPU) programing. LiDAR points are first projected onto a rasterized x–z plane so that sparse points are mapped into a series of regularly arranged small cells. Based on the height distribution of the LiDAR point, the ground cells are filtered out and a flag map is generated. Next, the ER-CCL algorithm is implemented on the label map generated from the flag map to mark individual clusters with unique labels. Finally, obstacle labeling results are inverse transformed from the x–z plane to 3D points to provide clustering results. For real-time 3D point cloud clustering, ER-CCL is accelerated by running it in parallel with the aid of GPU programming technology.

A robust segmentation framework for closely packed buildings from airborne LiDAR point clouds

ABSTRACT Urban villages (UVs) are commonly found in many Asian cities. These villages contain many closely packed buildings constructed decades ago without proper urban planning. There is a need for those buildings to be identified and put into statistics. In this paper, we present a segmentation framework that invokes multiple machine learning techniques and point cloud/image processing algorithms to segment individual closely packed buildings from large urban scenes. The presented framework consists of two major segmentation processes. The framework first filters out the non-ground objects from the point cloud, then it classified them by using the Random Forest classifier to isolate buildings from the entire scene. After that, the building point clouds will be segmented based on several building attribute analysis methods. This is followed by using the Random Sample Consensus (RANSAC) plane filtering method to expand the space between two closely packed buildings, so that the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering technique can be used to more accurately segment each individual building from the closely packed building areas. Two airborne Light Detection and Ranging (LiDAR) datasets collected in two different cities with some typical closely packed buildings were used to verify the proposed framework. The results show that the framework can effectively identify the closely packed buildings with unified structures from large airborne LiDAR datasets. The overall segmentation accuracy reaches 84% for the two datasets. The proposed framework can serve as a basis for analysis and segmentation of closely packed buildings with a more complicated structure.

Aerial Laser Scanning Data as a Source of Terrain Modeling in a Fluvial Environment: Biasing Factors of Terrain Height Accuracy.

Airborne light detection and ranging (LiDAR) scanning is a commonly used technology for representing the topographic terrain. As LiDAR point clouds include all surface features present in the terrain, one of the key elements for generating a digital terrain model (DTM) is the separation of the ground points. In this study, we intended to reveal the efficiency of different denoising approaches and an easy-to-use ground point classification technique in a floodplain with fluvial forms. We analyzed a point cloud from the perspective of the efficiency of noise reduction, parametrizing a ground point classifier (cloth simulation filter, CSF), interpolation methods and resolutions. Noise filtering resulted a wide range of point numbers in the models, and the number of points had moderate correlation with the mean accuracies (r = −0.65, p < 0.05), indicating that greater numbers of points had larger errors. The smallest differences belonged to the neighborhood-based noise filtering and the larger cloth size (5) and the smaller threshold value (0.2). The most accurate model was generated with the natural neighbor interpolation with the cloth size of 5 and the threshold of 0.2. These results can serve as a guide for researchers using point clouds when considering the steps of data preparation, classification, or interpolation in a flat terrain.

Visual analytics of 3D LiDAR point clouds in robotics operating systems

This paper presents visual analytics of 3D LiDAR point clouds in robotics operating system. In this study, experiment on Simultaneous Localization and Mapping (SLAM) using point cloud data derived from the Light Detection and Ranging (LiDAR) technology is conducted. We argue that one of the weaknesses of the SLAM algorithm is in the localization process of the landmarks. Existing algorithms such as Grid Mapping and Monte Carlo have limitations in dealing with 3D environment data that have led to less accurate estimation. Therefore, this research proposes the SLAM algorithm based on Real-Time Appearance-Based (RTAB) and makes use of the Red Green Blue (RGB) camera for visualisation. The algorithm was tested by using the map data that was collected and simulated on the Robot Operating System (ROS) in Linux environment. We present the results and demonstrates that the map produced by RTAB is better compared to its counterparts. In addition, the probability for the estimated location is improved which allows for better vehicle maneuverability.

Curved Buildings Reconstruction From Airborne LiDAR Data by Matching and Deforming Geometric Primitives

Airborne LiDAR (Light Detection and Ranging) data is widely applied in building reconstruction, with studies reporting success in typical buildings. However, the reconstruction of curved buildings remains an open research problem. To this end, we propose a new framework for curved building reconstruction via assembling and deforming geometric primitives. The input LiDAR point cloud are first converted into contours where individual buildings are identified. After recognizing geometric units (primitives) from building contours, we get initial models by matching basic geometric primitives to these primitives. To polish assembly models, we employ a warping field for model refinements. Specifically, an embedded deformation (ED) graph is constructed via downsampling the initial model. Then, the point-to-model displacements are minimized by adjusting node parameters in the ED graph based on our objective function. The presented framework is validated on several highly curved buildings collected by various LiDAR in different cities. The experimental results, as well as accuracy comparison, demonstrate the advantage and effectiveness of our method. {The new insight attributes to an efficient reconstruction manner.} Moreover, we prove that the primitive-based framework significantly reduces the data storage to 10-20 percent of classical mesh models.

Individual Tree Detection in a Eucalyptus Plantation Using Unmanned Aerial Vehicle (UAV)-LiDAR

The present study addresses the tree counting of a Eucalyptus plantation, the most widely planted hardwood in the world. Unmanned aerial vehicle (UAV) light detection and ranging (LiDAR) was used for the estimation of Eucalyptus trees. LiDAR-based estimation of Eucalyptus is a challenge due to the irregular shape and multiple trunks. To overcome this difficulty, the layer of the point cloud containing the stems was automatically classified and extracted according to the height thresholds, and those points were horizontally projected. Two different procedures were applied on these points. One is based on creating a buffer around each single point and combining the overlapping resulting polygons. The other one consists of a two-dimensional raster calculated from a kernel density estimation with an axis-aligned bivariate quartic kernel. Results were assessed against the manual interpretation of the LiDAR point cloud. Both methods yielded a detection rate (DR) of 103.7% and 113.6%, respectively. Results of the application of the local maxima filter to the canopy height model (CHM) intensely depends on the algorithm and the CHM pixel size. Additionally, the height of each tree was calculated from the CHM. Estimates of tree height produced from the CHM was sensitive to spatial resolution. A resolution of 2.0 m produced a R2 and a root mean square error (RMSE) of 0.99 m and 0.34 m, respectively. A finer resolution of 0.5 m produced a more accurate height estimation, with a R2 and a RMSE of 0.99 and 0.44 m, respectively. The quality of the results is a step toward precision forestry in eucalypt plantations.

Fast Planar Detection System Using a GPU-Based 3D Hough Transform for LiDAR Point Clouds

Plane extraction is regarded as a necessary function that supports judgment basis in many applications, including semantic digital map reconstruction and path planning for unmanned ground vehicles. Owing to the heterogeneous density and unstructured spatial distribution of three-dimensional (3D) point clouds collected by light detection and ranging (LiDAR), plane extraction from it is recently a significant challenge. This paper proposed a parallel 3D Hough transform algorithm to realize rapid and precise plane detection from 3D LiDAR point clouds. After transforming all the 3D points from a Cartesian coordinate system to a pre-defined 3D Hough space, the generated Hough space is rasterised into a series of arranged cells to store the resided point counts into individual cells. A 3D connected component labeling algorithm is developed to cluster the cells with high values in Hough space into several clusters. The peaks from these clusters are extracted so that the targeting planar surfaces are obtained in polar coordinates. Because the laser beams emitted by LiDAR sensor holds several fixed angles, the collected 3D point clouds distribute as several horizontal and parallel circles in plane surfaces. This kind of horizontal and parallel circles mislead plane detecting results from horizontal wall surfaces to parallel planes. For detecting accurate plane parameters, this paper adopts a fraction-to-fraction method to gradually transform raw point clouds into a series of sub Hough space buffers. In our proposed planar detection algorithm, a graphic processing unit (GPU) programming technology is applied to speed up the calculation of 3D Hough space updating and peaks searching.

Big Data Geospatial Processing for Massive Aerial LiDAR Datasets

For years, Light Detection and Ranging (LiDAR) technology has been considered as a challenge when it comes to developing efficient software to handle the extremely large volumes of data this surveying method is able to collect. In contexts such as this, big data technologies have been providing powerful solutions for distributed storage and computing. In this work, a big data approach on geospatial processing for massive aerial LiDAR point clouds is presented. By using Cassandra and Spark, our proposal is intended to support the execution of any kind of heavy time-consuming process; nonetheless, as an initial case of study, we have focused on fast ground-only rasters obtention to generate digital terrain models (DTMs) from massive LiDAR datasets. Filtered clouds obtained from the isolated processing of adjacent zones may exhibit errors located on the boundaries of the zones in the form of misclassified points. Usually, this type of error is corrected through manual or semi-automatic procedures. In this work, we also present an automated strategy for correcting errors of this type, improving the quality of the classification process and the DTMs obtained while minimizing user intervention. The autonomous nature of all computing stages, along with the low processing times achieved, opens the possibility of considering the system as a highly scalable service-oriented solution for on-demand DTM generation or any other geospatial process. Said solution would be a highly useful and unique service for many users in the LiDAR field, and one which could get near to real-time processing with appropriate computational resources.

Directionally constrained fully convolutional neural network for airborne LiDAR point cloud classification

Point cloud classification plays an important role in a wide range of airborne light detection and ranging (LiDAR) applications, such as topographic mapping, forest monitoring, power line detection, and road detection. However, due to the sensor noise, high redundancy, incompleteness, and complexity of airborne LiDAR systems, point cloud classification is challenging. Traditional point cloud classification methods mostly focus on the development of handcrafted point geometry features and employ machine learning-based classification models to conduct point classification. In recent years, the advances of deep learning models have caused researchers to shift their focus towards machine learning-based models, specifically deep neural networks, to classify airborne LiDAR point clouds. These learning-based methods start by transforming the unstructured 3D point sets to regular 2D representations, such as collections of feature images, and then employ a 2D CNN for point classification. Moreover, these methods usually need to calculate additional local geometry features, such as planarity, sphericity and roughness, to make use of the local structural information in the original 3D space. Nonetheless, the 3D to 2D conversion results in information loss. In this paper, we propose a directionally constrained fully convolutional neural network (D-FCN) that can take the original 3D coordinates and LiDAR intensity as input; thus, it can directly apply to unstructured 3D point clouds for semantic labeling. Specifically, we first introduce a novel directionally constrained point convolution (D-Conv) module to extract locally representative features of 3D point sets from the projected 2D receptive fields. To make full use of the orientation information of neighborhood points, the proposed D-Conv module performs convolution in an orientation-aware manner by using a directionally constrained nearest neighborhood search. Then, we design a multiscale fully convolutional neural network with downsampling and upsampling blocks to enable multiscale point feature learning. The proposed D-FCN model can therefore process input point cloud with arbitrary sizes and directly predict the semantic labels for all the input points in an end-to-end manner. Without involving additional geometry features as input, the proposed method demonstrates superior performance on the International Society for Photogrammetry and Remote Sensing (ISPRS) 3D labeling benchmark dataset. The results show that our model achieves a new state-of-the-art performance on powerline, car, and facade categories. Moreover, to demonstrate the generalization abilities of the proposed method, we conduct further experiments on the 2019 Data Fusion Contest Dataset. Our proposed method achieves superior performance than the comparing methods and accomplishes an overall accuracy of 95.6% and an average F1 score of 0.810.

Direct estimation of leaf area index of tropical forests using LiDAR point cloud

Light Detection and Ranging (LiDAR) remote sensing is a promising method for accurate estimation of various forest structural parameters. A ground based LiDAR scanner, terrestrial laser scanner (TLS) has been used for the reference field measurements for the estimation of structural parameters such as tree height, DBH across large area forests sites. Leaf Area Index (LAI) is important forest structural parameter which has recently been estimated from TLS point cloud by geometrical/statistical modeling with tree height and DBH as proxy variables. In this work we present a method for direct estimation of LAI of tropical forests from TLS point cloud proposing a new algorithmic approach named as Point Spatial Density (PSD) algorithm. As an intermediate processing, the proposed method involves filtering and 3D reconstruction of individual trees using super voxel and min-cut segmentation approaches. Reconstructed trees are then extracted and geometrically modelled for estimating the LAI by the proposed PSD algorithm. The proposed PSD algorithm establishes a relationship between TLS point density, point spacing, and height of trees. Based on the number of points and the point spacing, the value of LAI found has been observed to be varying in sparse and dense canopy tree stands. Validation of the results with reference LAI measurements obtained using a LAI meter (LiCOR LAI Meter 2200) indicates a consistent correlation (R2 = 0.96) between the estimated and reference LAI measurements. The results suggest that the LAI in dense heterogeneous forests could be accurately estimated by the proposed PSD algorithm based approach.

Automatic Delineation of Forest Patches in Highly Fragmented Landscapes Using Coloured Point Clouds

Accurate mapping of landscape features is key for natural resources management and planning. For this purpose, the use of high-resolution remote sensing data has become widespread and is increasingly freely available. However, mapping some target features, such as small forest patches, is still a challenge. Standard, easily replicable, and automatic methodologies to delineate such features are still missing. A common alternative to automated methods is manual delineation, but this is often too time and resource intensive. We developed a simple and automatic method from freely available aerial light detection and ranging (LiDAR) and aerial ortho-images that provide accurate land use mapping and overcome some of the aforementioned limitations. The input for the algorithm is a coloured point cloud, where multispectral information from the ortho-images is associated to each LiDAR point. From this, four-class segmentation and mapping were performed based on vegetation indices and the ground-elevation of the points. We tested the method in four areas in the north-western Iberian Peninsula and compared the results with existent cartography. The completeness and correctness of our algorithm ranging between 78% and 99% in most cases, and it allows for the delineation of very small patches that were previously underrepresented in the reference cartography.

EXTRACTION OF HOUSES FROM POINT CLOUD LIDAR: PROBLEMS AND CHALLENGE

Abstract. Although many efforts have been made on the extraction of houses from LiDAR (Light Detection and Ranging) and/or aerial imagery and/or their fusion, little investigation using co-registration between the orthoimage map and LiDAR on the basis of geodetic coordinates as element for house extraction. For this reason, this paper first overviews the advances of LiDAR and investigates the advantages and disadvantages of LiDAR system vs. traditional photogrammetry, and then indicates that LiDAR technology has not yet resolved all existing problems that traditional photogrammetry remained so far, such as texture information, LiDAR point cloud density. A comprehensive comparison in extraction of houses (feature information) from LiDAR data set and from aerial imagery are also presented. It has been widely accepted and admitted that full automation for extraction of houses (feature information in city area) from LiDAR point cloud has still been difficult. Therefore, this paper proposes a human-computer interaction operation for houses extraction through combination of LiDAR point cloud and the orthorectified high-resolution aerial imagery. The real data is utilized for validation of the proposed method.

Lattice-Constrained Stratified Sampling for Point Cloud Levels of Detail

In the last several years, airborne topographic light detection and ranging (LiDAR) has emerged as an important remote-sensing technology supporting a wide variety of applications. In that time, researchers have conducted studies to determine the optimal sampling densities required to support their respective applications. This natural progression of experimentation and analysis has resulted in several recommended sampling density stratifications for LiDAR point cloud products. Recognizing the need for point clouds of varying sample densities provides at least two motivations for creating level of details (LOD) for high-resolution point clouds. First, from the perspective of LiDAR data consumers, there is a desire to use the coarsest sampling that supports the application to reduce procurement costs, storage constraints, and processing times. Second, from the perspective of LiDAR data providers, there is a desire to collect data once at the highest supported fidelity to minimize recollection costs and redundancy in data holdings. In this article, we present an approach for generating point cloud LODs by constraining samples to regular discrete lattices to optimize the coverage of the volumes represented by each sample. We compare our approach to the two most common point cloud sampling methods: random sampling and rectangular lattice sampling. We discuss two approaches for representing the point cloud LODs. Finally, we propose an extension of our sampling approach for processing single-photon and Geiger-mode avalanche photodiode (GmAPD) LiDAR.

Tensor Tucker Decomposition based Geometry Compression on Three Dimensional LiDAR Point Cloud Image

Data Visualization in static images is still dynamically growing and changing with time in recent days. In visualization applications, memory, time and bandwidth are crucial issues when handling the high resolution three dimensional (3D) Light Detection and Ranging (LiDAR) data and they progressively demand efficient data compression strategies. This shortage is strongly motivating us to develop an efficient 3D point cloud image compression methodology. This work introduces an innovative lossless compression algorithm for a 3D point cloud image based on higher-order singular value decomposition (HOSVD) technique. This algorithm starts with the preprocessing method which removes the unreliable 3D points and then it combines the HOSVD together with the normalization, predictive coding followed by Run Length encoding to compress the HOSVD coefficients. This work accomplished lower mean square error (MSE), high (infinitive) Peak signal noise ratio (PSNR) to produce the lossless decompressed 3D point cloud image. The storage size has been reduced to one by fourth of its original 3D LiDAR point cloud image size.

Geometric calibration for LiDAR-camera system fusing 3D-2D and 3D-3D point correspondences.

Calibrating the extrinsic parameters on a system of 3D Light Detection And Ranging (LiDAR) and the monocular camera is a challenging task, because accurate 3D-2D or 3D-3D point correspondences are hard to establish from the sparse LiDAR point clouds in the calibration procedure. In this paper, we propose a geometric calibration method for estimating the extrinsic parameters of the LiDAR-camera system. In this method, a novel combination of planar boards with chessboard patterns and auxiliary calibration objects are proposed. The planar chessboard provides 3D-2D and 3D-3D point correspondences. Auxiliary calibration objects provide extra constraints for stable calibration results. After that, a novel geometric optimization framework is proposed to utilize these point correspondences, thus leading calibration results robust to LiDAR sensor noise. Besides, we contribute an automatic approach to extract point clouds of calibration objects. In the experiments, our method has a superior performance over state-of-the-art calibration methods. Furthermore, we verify our method by computing depth map and improvements can also be found. These results demonstrate that our method performance on the LiDAR-camera system is applicable for future advanced visual applications.

Reclaimed Area Land Cover Mapping Using Sentinel-2 Imagery and LiDAR Point Clouds

This paper investigates the possibility of using fusion Sentinel-2 imageries (2016, ESA) and light detection and ranging (LiDAR) point clouds for the automation of land cover mapping with a primary focus on detecting and monitoring afforested areas and deriving precise information about the spatial (2D and 3D) characteristics of vegetation for reclaimed areas. The study was carried out for reclaimed areas – two former sulfur mines located in Southeast Poland, namely, Jeziórko, where 216.5 ha of afforested area was reclaimed after borehole exploitation, and Machów, where 871.7 ha of dump area was reclaimed after open cast strip mining. The current land use and land cover (LULC) classes at the Machów and Jeziórko former sulfur mines are derived based on Sentinel-2 image processing, and confirmed the applied type of reclamation for both analysed areas. The following LULC classes showed a significant spatial range: broad-leaved forest, coniferous forest, and transitional woodland shrub. The progress of afforested areas, not only in terms of the occupied area, but also in terms of the growth of trees and shrubs, was confirmed. The results of the study showed differences in vegetation parameters, namely, height and canopy cover. Various stages of vegetation growth were also observed. This indicates an ongoing process of vegetation development, as an effect of the reclamation treatment for these areas.