Towards a point cloud understanding framework for forest scene semantic segmentation across forest types and sensor platforms

Remote Sensing of Environment, Год журнала: 2025, Номер 318, С. 114591 - 114591

Опубликована: Янв. 15, 2025

Язык: Английский

---

A clustering-based automatic registration of UAV and terrestrial LiDAR forest point clouds

Computers and Electronics in Agriculture, Год журнала: 2024, Номер 217, С. 108648 - 108648

Опубликована: Янв. 19, 2024

Язык: Английский

---

The full-field displacement intelligent measurement of retaining structures using UAV and 3D reconstruction

Measurement, Год журнала: 2024, Номер 227, С. 114311 - 114311

Опубликована: Фев. 11, 2024

Язык: Английский

---

Harnessing Artificial Intelligence, Machine Learning and Deep Learning for Sustainable Forestry Management and Conservation: Transformative Potential and Future Perspectives

Plants, Год журнала: 2025, Номер 14(7), С. 998 - 998

Опубликована: Март 22, 2025

Plants serve as the basis for ecosystems and provide a wide range of essential ecological, environmental, economic benefits. However, forest plants other systems are constantly threatened by degradation extinction, mainly due to misuse exhaustion. Therefore, sustainable management (SFM) is paramount, especially in wake global climate change challenges. SFM ensures continued provision forests both present future generations. In practice, faces challenges balancing use conservation forests. This review discusses transformative potential artificial intelligence (AI), machine learning, deep learning (DL) technologies management. It summarizes current research technological improvements implemented using AI, discussing their applications, such predictive analytics modeling techniques that enable accurate forecasting dynamics carbon sequestration, species distribution, ecosystem conditions. Additionally, it explores how AI-powered decision support facilitate adaptive strategies integrating real-time data form images or videos. The manuscript also highlights limitations incurred ML, DL combating management, providing acceptable solutions these problems. concludes perspectives immense modernizing SFM. Nonetheless, great deal has already shed much light on this topic, bridges knowledge gap.

Язык: Английский

---

An Improved RANSAC-ICP Method for Registration of SLAM and UAV-LiDAR Point Cloud at Plot Scale

Forests, Год журнала: 2024, Номер 15(6), С. 893 - 893

Опубликована: Май 21, 2024

Simultaneous Localization and Mapping (SLAM) using LiDAR technology can acquire the point cloud below tree canopy efficiently in real time, Unmanned Aerial Vehicle (UAV-LiDAR) derive of canopy. By registering them, complete 3D structural information trees be obtained for forest inventory. To this end, an improved RANSAC-ICP algorithm registration SLAM UAV-LiDAR at plot scale is proposed study. Firstly, features are extracted transformed into 33-dimensional feature vectors by descriptor FPFH, corresponding pairs determined bidirectional matching. Then, RANSAC employed to compute transformation matrix based on reduced set points coarse cloud. Finally, iterative closest used iterate achieve precise The validated both coniferous broadleaf datasets, with average mean absolute distance (MAD) 11.332 cm dataset 6.150 dataset. experimental results show that method study effectively applied alignment multi-platform clouds.

Язык: Английский

---

Use of a Consumer-Grade UAV Laser Scanner to Identify Trees and Estimate Key Tree Attributes across a Point Density Range

Forests, Год журнала: 2024, Номер 15(6), С. 899 - 899

Опубликована: Май 22, 2024

The management of plantation forests using precision forestry requires advanced inventory methods. Unmanned aerial vehicle laser scanning (ULS) offers a cost-effective approach to accurately estimate forest structural attributes at both plot and individual tree levels. We examined the utility ULS data collected from radiata pine stand for detection prediction diameter breast height (DBH) stem volume, thinned 13-point densities (ranging 10–12,200 points/m2). These datasets were created DTM with highest pulse density DTMs that used native decimated point clouds. Models DBH constructed partial least squares (PLS) random (RF) seven classes metrics characterized horizontal vertical structure canopy. Individual segmentation was consistently accurate across insensitive type (F1 scores > 0.96). Predictions PLS models more than RF accuracy type. Using DTMs, estimation had lowest RMSE 1.624 cm (R2 0.756) 12,200 points/m2. Stem volume predictions made 0.0418 m3 0.792) values remained relatively stable between 750 400 points/m2, reductions in occurring as declined below this threshold. Overall, these findings have significant implications, particularly precise level. They demonstrate potential sensors rapid frequent assessment, thereby enhancing application light ranging (LiDAR) technology management.

Язык: Английский

---

Estimation of Rubber Plantation Biomass Based on Variable Optimization from Sentinel-2 Remote Sensing Imagery

Forests, Год журнала: 2024, Номер 15(6), С. 900 - 900

Опубликована: Май 22, 2024

The rapid, accurate, and non-destructive estimation of rubber plantation aboveground biomass (AGB) is essential for producers to forecast yield carbon storage. To enhance the accuracy, an increasing number remote sensing variables are incorporated into development multi-parameter models, which makes its practical application potential impact on predictive precision challenging due inclusion non-essential or redundant variables. Therefore, this study systematically evaluated performance different parameter combinations derived from Sentinel-2 imagery, using variable optimization approaches with four machine learning algorithms (Random Forest Regression, RF; XGBoost XGBR; K Nearest Neighbor KNNR; Support Vector SVR) AGB plantations. results indicate that RF achieved best accuracy (R2 = 0.86, RMSE 15.77 Mg/ha) predicting when combined Boruta-selected variables, outperforming other (variable obtained based importance ranking, univariate combinations, multivariate combinations). Our research findings suggest consideration parameter-optimized advantageous improving forest biophysical parameters, utilizing a large parameters estimation.

Язык: Английский

---

An Individual Tree Detection and Segmentation Method from TLS and MLS Point Clouds Based on Improved Seed Points

Forests, Год журнала: 2024, Номер 15(7), С. 1083 - 1083

Опубликована: Июнь 22, 2024

Individual Tree Detection and Segmentation (ITDS) is a key step in accurately extracting forest structural parameters from LiDAR (Light Ranging) data. However, most ITDS algorithms face challenges with over-segmentation, under-segmentation, the omission of small trees high-density forests. In this study, we developed bottom–up framework for based on seed points. The proposed method density-based spatial clustering applications noise (DBSCAN) to initially detect trunks filter clusters by set threshold. Then, K-Nearest Neighbor (KNN) algorithm used reclassify non-core clustered point cloud after threshold filtering. Furthermore, Random Sample Consensus (RANSAC) cylinder fitting correct trunk detection results. Finally, calculate centroid clouds as points achieve individual tree segmentation (ITS). paper, use terrestrial laser scanning (TLS) data natural forests Germany mobile (MLS) planted China explore effects accuracy ITS methods; then evaluate efficiency three aspects: detection, overall segmentation. We show following: (1) addresses issues missing misrecognition DBSCAN detection. Compared using directly, recall (r), precision (p), F-score (F) increased 6.0%, 6.5%, 0.07, respectively; (2) significantly improved (3) achieved r, p, F 95.2%, 97.4%, 0.96, respectively. This work demonstrates excellent able segment under tall trees.

Язык: Английский

---

UAV-derived photogrammetric point clouds and multispectral indices for fuel estimation in Mediterranean forests

Remote Sensing Applications Society and Environment, Год журнала: 2023, Номер 31, С. 100997 - 100997

Опубликована: Май 25, 2023

Sensors attached to unmanned aerial vehicles (UAVs) allow estimating a large number of forest attributes related fuels. This study assesses photogrammetric point clouds and multispectral indices obtained from fixed-wing UAV for the classification Prometheus fuel types in 82 plots Aragón (NE Spain). Images captured by an RGB camera sensor allowed generating high density (RGB: 3000 points/m2; multispectral: 85 points/m2), which were normalized using alternatively Digital Elevation Model (DEM) 0.5, 1, 2 m resolution. A set structural textural variables derived cloud heights, latter, gray-level co-occurrence matrix (GLCM) approach was used. Multispectral images also used create seven spectral vegetation indices. The most relevant structural, textural, introduce into models selected Dunn's test, included: height at 50th percentile, coefficient variation percentage returns above 4 m, mean dissimilarity, Green Chlorophyll Index. Three different data samples introduced models: i) (RGB sample); ii) (MS iii) plus variable (integrated sample). After comparing three machine learning techniques (Random Forest, Linear Radial Support Vector Machine), best results with Random Forest k-fold cross-validation (k-10) integrated sample 0.5 DEM resolution (overall accuracy = 71%). successfully identified main fire carriers (i.e., shrubs or trees) confusions mainly located within same dominant stratum, especially 3 6. These demonstrate ability imagery classify fuels Mediterranean environments when are combined.

Язык: Английский

---

A Feature-Level Point Cloud Fusion Method for Timber Volume of Forest Stands Estimation

Remote Sensing, Год журнала: 2023, Номер 15(12), С. 2995 - 2995

Опубликована: Июнь 8, 2023

Accurate diameter at breast height (DBH) and tree (H) information can be acquired through terrestrial laser scanning (TLS) airborne LiDAR scanner (ALS) point cloud, respectively. To utilize these two features simultaneously but avoid the difficulties of cloud fusion, such as technical complexity time-consuming laborious efforts, a feature-level fusion method (FFATTe) is proposed in this paper. Firstly, TLS ALS data plot are georeferenced by differential global navigation positioning system (DGNSS) technology. Secondly, processing feature extraction performed for to form datasets, Thirdly, from different sources realized spatial join according trunk location obtained ALS, that is, tally implemented plot. Finally, individual parameters optimized based on results fed into binary volume model estimate total (TVS) large area (whole study area). The show using DGNSS RTK/PPK technology achieve coarse registration (mean distance ≈ 40 cm), which meets accuracy requirements fusion. By data, achieved quickly accurately FFATTe achieves high (with error 3.09%) due its advantages combining simple way, it has strong operability when acquiring TVS over areas.

Язык: Английский

---