Estimating Forest Structure from UAV-Mounted LiDAR Point Cloud Using Machine Learning

A Comprehensive Review of LiDAR Applications in Crop Management for Precision Agriculture

Precision agriculture has revolutionized crop management and agricultural production, with LiDAR technology attracting significant interest among various technological advancements. This extensive review examines the various applications of LiDAR in precision agriculture, with a particular emphasis on its function in crop cultivation and harvests. The introduction provides an overview of precision agriculture, highlighting the need for effective agricultural management and the growing significance of LiDAR technology. The prospective advantages of LiDAR for increasing productivity, optimizing resource utilization, managing crop diseases and pesticides, and reducing environmental impact are discussed. The introduction comprehensively covers LiDAR technology in precision agriculture, detailing airborne, terrestrial, and mobile systems along with their specialized applications in the field. After that, the paper reviews the several uses of LiDAR in agricultural cultivation, including crop growth and yield estimate, disease detection, weed control, and plant health evaluation. The use of LiDAR for soil analysis and management, including soil mapping and categorization and the measurement of moisture content and nutrient levels, is reviewed. Additionally, the article examines how LiDAR is used for harvesting crops, including its use in autonomous harvesting systems, post-harvest quality evaluation, and the prediction of crop maturity and yield. Future perspectives, emergent trends, and innovative developments in LiDAR technology for precision agriculture are discussed, along with the critical challenges and research gaps that must be filled. The review concludes by emphasizing potential solutions and future directions for maximizing LiDAR's potential in precision agriculture. This in-depth review of the uses of LiDAR gives helpful insights for academics, practitioners, and stakeholders interested in using this technology for effective and environmentally friendly crop management, which will eventually contribute to the development of precision agricultural methods.

Design of Lidar Receiving Optical System with Large FoV and High Concentration of Light to Resist Background Light Interference

Lidar has the advantages of high accuracy, high resolution, and is not affected by sunlight. It has been widely used in many fields, such as autonomous driving, remote sensing detection, and intelligent robots. However, the current lidar detection system belongs to weak signal detection and generally uses avalanche photoelectric detector units as detectors. Limited by the current technology, the photosensitive surface is small, the receiving field of view is limited, and it is easy to cause false alarms due to background light. This paper proposes a method based on a combination of image-side telecentric lenses, microlens arrays, and interference filters. The small-area element detector achieves the high-concentration reception of echo beams in a large field of view while overcoming the interference of ambient background light. The image-side telecentric lens realizes that the center lines of the echo beams at different angles are parallel to the central axis, and the focus points converge on the same focal plane. The microlens array collimates the converged light beams one by one into parallel light beams. Finally, a high-quality aspherical focusing lens is used to focus the light on the small-area element detector to achieve high-concentration light reception over a large field of view. The system achieves a receiving field of view greater than 40° for a photosensitive surface detector with a diameter of 75 μm and is resistant to background light interference.

From UAV to PlanetScope: Upscaling fractional cover of an invasive species Rosa rugosa

Invasive plant species pose a direct threat to biodiversity and ecosystem services. Among these, Rosa rugosa has had a severe impact on Baltic coastal ecosystems in recent decades. Accurate mapping and monitoring tools are essential to quantify the location and spatial extent of invasive plant species to support eradication programs. In this paper we combined RGB images obtained using an Unoccupied Aerial Vehicle, with multispectral PlanetScope images to map the extent of R. rugosa at seven locations along the Estonian coastline. We used RGB-based vegetation indices and 3D canopy metrics in combination with a random forest algorithm to map R. rugosa thickets, obtaining high mapping accuracies (Sensitivity = 0.92, specificity = 0.96). We then used the R. rugosa presence/absence maps as a training dataset to predict the fractional cover based on multispectral vegetation indices derived from the PlanetScope constellation and an Extreme Gradient Boosting algorithm (XGBoost). The XGBoost algorithm yielded high fractional cover prediction accuracies (RMSE = 0.11, R² = 0.70). An in-depth accuracy assessment based on site-specific validations revealed notable differences in accuracy between study sites (highest R² = 0.74, lowest R² = 0.03). We attribute these differences to the various stages of R. rugosa invasion and the density of thickets. In conclusion, the combination of RGB UAV images and multispectral PlanetScope images is a cost-effective method to map R. rugosa in highly heterogeneous coastal ecosystems. We propose this approach as a valuable tool to extend the highly local geographical scope of UAV assessments into wider areas and regional evaluations.

Machine learning assisted remote forestry health assessment: a comprehensive state of the art review

Forests are suffering water stress due to climate change; in some parts of the globe, forests are being exposed to the highest temperatures historically recorded. Machine learning techniques combined with robotic platforms and artificial vision systems have been used to provide remote monitoring of the health of the forest, including moisture content, chlorophyll, and nitrogen estimation, forest canopy, and forest degradation, among others. However, artificial intelligence techniques evolve fast associated with the computational resources; data acquisition, and processing change accordingly. This article is aimed at gathering the latest developments in remote monitoring of the health of the forests, with special emphasis on the most important vegetation parameters (structural and morphological), using machine learning techniques. The analysis presented here gathered 108 articles from the last 5 years, and we conclude by showing the newest developments in AI tools that might be used in the near future.

The Influence of Data Density and Integration on Forest Canopy Cover Mapping Using Sentinel-1 and Sentinel-2 Time Series in Mediterranean Oak Forests

Forest canopy cover (FCC) is one of the most important forest inventory parameters and plays a critical role in evaluating forest functions. This study examines the potential of integrating Sentinel-1 (S-1) and Sentinel-2 (S-2) data to map FCC in the heterogeneous Mediterranean oak forests of western Iran in different data densities (one-year datasets vs. three-year datasets). This study used very high-resolution satellite images from Google Earth, gridded points, and field inventory plots to generate a reference dataset. Based on it, four FCC classes were defined, namely non-forest, sparse forest (FCC = 1–30%), medium-density forest (FCC = 31–60%), and dense forest (FCC > 60%). In this study, three machine learning (ML) models, including Random Forest (RF), Support Vector Machine (SVM), and Classification and Regression Tree (CART), were used in the Google Earth Engine and their performance was compared for classification. Results showed that the SVM produced the highest accuracy on FCC mapping. The three-year time series increased the ability of all ML models to classify FCC classes, in particular the sparse forest class, which was not distinguished well by the one-year dataset. Class-level accuracy assessment results showed a remarkable increase in F-1 scores for sparse forest classification by integrating S-1 and S-2 (10.4% to 18.2% increased for the CART and SVM ML models, respectively). In conclusion, the synergetic use of S-1 and S-2 spectral temporal metrics improved the classification accuracy compared to that obtained using only S-2. The study relied on open data and freely available tools and can be integrated into national monitoring systems of FCC in Mediterranean oak forests of Iran and neighboring countries with similar forest attributes.

A Hybrid Framework Combining Data-Driven and Catenary-Based Methods for Wide-Area Powerline Sag Estimation

This paper is concerned with the airborne-laser-data-based sag estimation for wide-area transmission lines. A systematic data processing framework is established for multi-source data collected from power lines, which is applicable to various operating conditions. Subsequently, a k-means-based clustering approach is employed to handle the spatial heterogeneity and sparsity of powerline corridor data after comprehensive performance comparisons. Furthermore, a hybrid model of the catenary and XGBoost (HMCX) method is proposed for sag estimation, which improves the accuracy of sag estimation by integrating the adaptability of catenary and the sparsity awareness of XGBoost. Finally, the effectiveness of HMCX is verified by using power data from 116 actual lines.

The Use of UAV-Acquired Multiband Images for Detecting Rockfall-Induced Injuries at Tree Crown Level

In this paper, we present an identification of rockfall-injured trees based on multiband images obtained by an unmanned aerial vehicle (UAV). A survey with a multispectral camera was performed on three rockfall sites with versatile tree species (Fagus sylvatica L., Larix decidua Mill., Pinus sylvestris L., Picea abies (L.) Karsten, and Abies alba Mill.) and with different characterizations of rockfalls and rockfall-induced injuries. At one site, rockfall injuries were induced in the same year as the survey. At the second site, they were induced one year after the initial injuries, and at the third site, they were induced six years after the first injuries. At one site, surveys were performed three years in a row. Multiband images were used to extract different vegetation indices (VIs) at the tree crown level and were further studied to see which VIs can identify the injured trees and how successfully. A total of 14 VIs were considered, including individual multispectral bands (green, red, red edge, and near-infrared) by using regression models to differentiate between the injured and uninjured groups for a single year and for three consecutive years. The same model was also used for VI differentiations among the recorded injury groups and size of the injuries. The identification of injured trees based on VIs was possible at the sites where rockfall injuries were induced at least one year before the UAV survey, and they could still be identifiable six years after the initial injuries. At the site where injuries were induced only four months before the UAV survey, the identification of injured trees was not possible. VIs that could explain the largest variability (R2 > 0.3) between injured and uninjured trees were: inverse ratio index (IRVI), green–red vegetation index (GRVI), normalized difference vegetation index (NDVI), normalized ratio index (NRVI), and ratio vegetation index (RVI). RVI was the most successful, explaining 40% of the variance at two sites. R2 values only increased by a few percentages (up to 10%) when the VIs of injured trees were observed over a period of three years and mostly did not change significantly, thus not indicating if the vitality of the trees increased or decreased. Differentiation among the injured groups did not show promising results, while, on the other hand, there was a strong correlation between the VI values (RVI) and the size of the injury according to the basal area of the trees (so-called injury index). Both in the case of broadleaves and conifers at two sites, the R2 achieved a value of 0.82. The presented results indicate that the UAV-acquired multiband images at the tree crown level can be used for surveying rockfall protection forests in order to monitor their vitality, which is crucial for maintaining the protective effect through time and space.

Exploring the Potential of UAV LiDAR Data for Trunk Point Extraction and Direct DBH Measurement

The accurate measurement of diameter at breast height (DBH) is one of the essential tasks for biomass estimation at an individual tree scale. This paper aims to explore the potential of unmanned aerial vehicle (UAV) based light detection and ranging (LiDAR) for trunk point extraction and direct DBH measurement. First, the trunk point cloud for each tree is extracted based on UAV LiDAR data by the multiscale cylindrical detection method. Then, the DBH is directly measured from the point cloud via the multiscale ring fitting. Lastly, we analyze the influence of scanning angle and mode on trunk point extraction and DBH measurement. The results show that the proposed method can obtain high accuracy of trunk point extraction and DBH measurement with real (R2 = 0.708) and simulated (R2 = 0.882) UAV LiDAR data. It proves that the UAV LiDAR data is feasible to directly measure the DBH. The highest accuracy was obtained with the scanning angles ranging from 50 to 65 degrees. Additionally, as the number of routes increases, the accuracy increases. This paper demonstrates that the UAV LiDAR can be used to directly measure the DBH, providing the scientific guidance for UAV path planning and LiDAR scanning design.

A Method of Curve Reconstruction Based on Point Cloud Clustering and PCA

In many application fields (closed curve noise data reconstruction, time series data fitting, image edge smoothing, skeleton extraction, etc.), curve reconstruction based on noise data has always been a popular but challenging problem. In a single domain, there are many methods for curve reconstruction of noise data, but a method suitable for multi-domain curve reconstruction has received much less attention in the literature. More importantly, the existing methods have shortcomings in time consumption when dealing with large data and high-density point cloud curve reconstruction. For this reason, we hope to propose a curve fitting algorithm suitable for many fields and low time consumption. In this paper, a curve reconstruction method based on clustering and point cloud principal component analysis is proposed. Firstly, the point cloud is clustered by the K++ means algorithm. Secondly, a denoising method based on point cloud principal component analysis is proposed to obtain the interpolation nodes of curve subdivision. Finally, the fitting curve is obtained by the parametric curve subdivision method. Comparative experiments show that our method is superior to the classical fitting method in terms of time consumption and effect. In addition, our method is not constrained by the shape of the point cloud, and can play a role in time series data, image thinning and edge smoothing.

Denoising for 3D Point Cloud Based on Regularization of a Statistical Low-Dimensional Manifold

A point cloud obtained by stereo matching algorithm or three-dimensional (3D) scanner generally contains much complex noise, which will affect the accuracy of subsequent surface reconstruction or visualization processing. To eliminate the complex noise, a new regularization algorithm for denoising was proposed. In view of the fact that 3D point clouds have low-dimensional structures, a statistical low-dimensional manifold (SLDM) model was established. By regularizing its dimensions, the denoising problem of the point cloud was expressed as an optimization problem based on the geometric constraints of the regularization term of the manifold. A low-dimensional smooth manifold model was constructed by discrete sampling, and solved by means of a statistical method and an alternating iterative method. The performance of the denoising algorithm was quantitatively evaluated from three aspects, i.e., the signal-to-noise ratio (SNR), mean square error (MSE) and structural similarity (SSIM). Analysis and comparison of performance showed that compared with the algebraic point-set surface (APSS), non-local denoising (NLD) and feature graph learning (FGL) algorithms, the mean SNR of the point cloud denoised using the proposed method increased by 1.22 DB, 1.81 DB and 1.20 DB, respectively, its mean MSE decreased by 0.096, 0.086 and 0.076, respectively, and its mean SSIM decreased by 0.023, 0.022 and 0.020, respectively, which shows that the proposed method is more effective in eliminating Gaussian noise and Laplace noise in common point clouds. The application cases showed that the proposed algorithm can retain the geometric feature information of point clouds while eliminating complex noise.

Evaluating the Use of Lidar to Discern Snag Characteristics Important for Wildlife

Standing dead trees (known as snags) are historically difficult to map and model using airborne laser scanning (ALS), or lidar. Specific snag characteristics are important for wildlife; for instance, a larger snag with a broken top can serve as a nesting platform for raptors. The objective of this study was to evaluate whether characteristics such as top intactness could be inferred from discrete-return ALS data. We collected structural information for 198 snags in closed-canopy conifer forest plots in Idaho. We selected 13 lidar metrics within 5 m diameter point clouds to serve as predictor variables in random forest (RF) models to classify snags into four groups by size (small (<40 cm diameter) or large (≥40 cm diameter)) and intactness (intact or broken top) across multiple iterations. We conducted these models first with all snags combined, and then ran the same models with only small or large snags. Overall accuracies were highest in RF models with large snags only (77%), but kappa statistics for all models were low (0.29–0.49). ALS data alone were not sufficient to identify top intactness for large snags; future studies combining ALS data with other remotely sensed data to improve classification of snag characteristics important for wildlife is encouraged.

Identifying Forest Structural Types along an Aridity Gradient in Peninsular Spain: Integrating Low-Density LiDAR, Forest Inventory, and Aridity Index

Forest structure is a key driver of forest functional processes. The characterization of forest structure across spatiotemporal scales is essential for forest monitoring and management. LiDAR data have proven particularly useful for cost-effectively estimating forest structural attributes. This paper evaluates the ability of combined forest inventory data and low-density discrete return airborne LiDAR data to discriminate main forest structural types in the Mediterranean-temperate transition ecotone. Firstly, we used six structural variables from the Spanish National Forest Inventory (SNFI) and an aridity index in a k-medoids algorithm to define the forest structural types. These variables were calculated for 2770 SNFI plots. We identified the main species for each structural type using the SNFI. Secondly, we developed a Random Forest model to predict the spatial distribution of structural types and create wall-to-wall maps from LiDAR data. The k-medoids clustering algorithm enabled the identification of four clusters of forest structures. A total of six out of forty-one potential LiDAR metrics were utilized in our Random Forest, after evaluating their importance in the Random Forest model. Selected metrics were, in decreasing order of importance, the percentage of all returns above 2 m, mean height of the canopy profile, the difference between the 90th and 50th height percentiles, the area under the canopy curve, and the 5th and the 95th percentile of the return heights. The model yielded an overall accuracy of 64.18%. The producer’s accuracy ranged between 36.11% and 88.93%. Our results confirm the potential of this approximation for the continuous monitoring of forest structures, which is key to guiding forest management in this region.

An Occupancy Mapping Method Based on K-Nearest Neighbours

OctoMap is an efficient probabilistic mapping framework to build occupancy maps from point clouds, representing 3D environments with cubic nodes in the octree. However, the map update policy in OctoMap has limitations. All the nodes containing points will be assigned with the same probability regardless of the points being noise, and the probability of one such node can only be increased with a single measurement. In addition, potentially occupied nodes with points inside but traversed by rays cast from the sensor to endpoints will be marked as free. To overcome these limitations in OctoMap, the current work presents a mapping method using the context of neighbouring points to update nodes containing points, with occupancy information of a point represented by the average distance from a point to its k-Nearest Neighbours. A relationship between the distance and the change in probability is defined with the Cumulative Density Function of average distances, potentially decreasing the probability of a node despite points being present inside. Experiments are conducted on 20 data sets to compare the proposed method with OctoMap. Results show that our method can achieve up to 10% improvement over the optimal performance of OctoMap.

Towards Amazon Forest Restoration: Automatic Detection of Species from UAV Imagery

Precise assessments of forest species’ composition help analyze biodiversity patterns, estimate wood stocks, and improve carbon stock estimates. Therefore, the objective of this work was to evaluate the use of high-resolution images obtained from Unmanned Aerial Vehicle (UAV) for the identification of forest species in areas of forest regeneration in the Amazon. For this purpose, convolutional neural networks (CNN) were trained using the Keras–Tensorflow package with the faster_rcnn_inception_v2_pets model. Samples of six forest species were used to train CNN. From these, attempts were made with the number of thresholds, which is the cutoff value of the function; any value below this output is considered 0, and values above are treated as an output 1; that is, values above the value stipulated in the Threshold are considered as identified species. The results showed that the reduction in the threshold decreases the accuracy of identification, as well as the overlap of the polygons of species identification. However, in comparison with the data collected in the field, it was observed that there exists a high correlation between the trees identified by the CNN and those observed in the plots. The statistical metrics used to validate the classification results showed that CNN are able to identify species with accuracy above 90%. Based on our results, which demonstrate good accuracy and precision in the identification of species, we conclude that convolutional neural networks are an effective tool in classifying objects from UAV images.

Under-Canopy UAV Laser Scanning Providing Canopy Height and Stem Volume Accurately

The automation of forest field reference data collection has been an intensive research objective for laser scanning scientists ever since the invention of terrestrial laser scanning more than two decades ago. In this study, we demonstrated that an under-canopy UAV laser scanning system utilizing a rotating laser scanner can alone provide accurate estimates of canopy height and stem volume for the majority of trees in a boreal forest. We mounted a rotating laser scanner based on a Velodyne VLP-16 sensor onboard a manually piloted UAV. The UAV was commanded with the help of a live video feed from the onboard camera. Since the system was based on a rotating laser scanner providing varying view angles, all important elements such as treetops, branches, trunks, and ground could be recorded with laser hits. In an experiment including two different forest structures, namely sparse and obstructed canopy, we showed that our system can measure the heights of individual trees with a bias of −20 cm and a standard error of 40 cm in the sparse forest and with a bias of −65 cm and a standard error of 1 m in the obstructed forest. The accuracy of the obtained tree height estimates was equivalent to airborne above-canopy UAV surveys conducted in similar forest conditions or even at the same sites. The higher underestimation and higher inaccuracy in the obstructed site can be attributed to three trees with a height exceeding 25 m and the reduced point density of these tree tops due to occlusion and the limited ranging capacity of the scanner. Additionally, we used our system to estimate the stem volumes of individual trees with a standard error at the level of 10%. This level of error is equivalent to the error obtained when merging above-canopy UAV laser scanner data with terrestrial point cloud data. The results show that we do not necessarily need a combination of terrestrial point clouds and point clouds collected using above-canopy UAV systems in order to accurately estimate the heights and the volumes of individual trees in reference data collection.

Path Planning Method for UAVs Based on Constrained Polygonal Space and an Extremely Sparse Waypoint Graph

Finding an optimal/quasi-optimal path for Unmanned Aerial Vehicles (UAVs) utilizing full map information yields time performance degradation in large and complex three-dimensional (3D) urban environments populated by various obstacles. A major portion of the computing time is usually wasted on modeling and exploration of spaces that have a very low possibility of providing optimal/sub-optimal paths. However, computing time can be significantly reduced by searching for paths solely in the spaces that have the highest priority of providing an optimal/sub-optimal path. Many Path Planning (PP) techniques have been proposed, but a majority of the existing techniques equally evaluate many spaces of the maps, including unlikely ones, thereby creating time performance issues. Ignoring high-probability spaces and instead exploring too many spaces on maps while searching for a path yields extensive computing-time overhead. This paper presents a new PP method that finds optimal/quasi-optimal and safe (e.g., collision-free) working paths for UAVs in a 3D urban environment encompassing substantial obstacles. By using Constrained Polygonal Space (CPS) and an Extremely Sparse Waypoint Graph (ESWG) while searching for a path, the proposed PP method significantly lowers pathfinding time complexity without degrading the length of the path by much. We suggest an intelligent method exploiting obstacle geometry information to constrain the search space in a 3D polygon form from which a quasi-optimal flyable path can be found quickly. Furthermore, we perform task modeling with an ESWG using as few nodes and edges from the CPS as possible, and we find an abstract path that is subsequently improved. The results achieved from extensive experiments, and comparison with prior methods certify the efficacy of the proposed method and verify the above assertions.

Estimating Canopy Density Parameters Time-Series for Winter Wheat Using UAS Mounted LiDAR

Monitoring of canopy density with related metrics such as leaf area index (LAI) makes a significant contribution to understanding and predicting processes in the soil–plant–atmosphere system and to indicating crop health and potential yield for farm management. Remote sensing methods using optical sensors that rely on spectral reflectance to calculate LAI have become more mainstream due to easy entry and availability. Methods with vegetation indices (VI) based on multispectral reflectance data essentially measure the green area index (GAI) or response to chlorophyll content of the canopy surface and not the entire aboveground biomass that may be present from non-green elements that are key to fully assessing the carbon budget. Methods with light detection and ranging (LiDAR) have started to emerge using gap fraction (GF) to estimate the plant area index (PAI) based on canopy density. These LiDAR methods have the main advantage of being sensitive to both green and non-green plant elements. They have primarily been applied to forest cover with manned airborne LiDAR systems (ALS) and have yet to be used extensively with crops such as winter wheat using LiDAR on unmanned aircraft systems (UAS). This study contributes to a better understanding of the potential of LiDAR as a tool to estimate canopy structure in precision farming. The LiDAR method proved to have a high to moderate correlation in spatial variation to the multispectral method. The LiDAR-derived PAI values closely resemble the SunScan Ceptometer GAI ground measurements taken early in the growing season before major stages of senescence. Later in the growing season, when the canopy density was at its highest, a possible overestimation may have occurred. This was most likely due to the chosen flight parameters not providing the best depictions of canopy density with consideration of the LiDAR’s perspective, as the ground-based destructive measurements provided lower values of PAI. Additionally, a distinction between total LiDAR-derived PAI, multispectral-derived GAI, and brown area index (BAI) is made to show how the active and passive optical sensor methods used in this study can complement each other throughout the growing season.