Evaluation of Rapeseed Leave Segmentation Accuracy Using Binocular Stereo Vision 3D Point Clouds

Agronomy, Journal Year: 2025, Volume and Issue: 15(1), P. 245 - 245

Published: Jan. 20, 2025

Point cloud segmentation is necessary for obtaining highly precise morphological traits in plant phenotyping. Although a huge development has occurred point segmentation, the of clouds from complex leaves still remains challenging. Rapeseed are critical cultivation and breeding, yet traditional two-dimensional imaging susceptible to reduced accuracy due occlusions between plants. The current study proposes use binocular stereo-vision technology obtain three-dimensional (3D) rapeseed at seedling bolting stages. were colorized based on elevation values order better process 3D data extract phenotypic parameters. Denoising methods selected source classification noise. However, ground clouds, we combined plane fitting with pass-through filtering denoising, while statistical was used denoising outliers generated during scanning. We found that, stage rapeseed, region-growing method helpful finding suitable parameter thresholds leaf Locally Convex Connected Patches (LCCP) clustering stage. Furthermore, results show that combining effectively removes noise, successfully denoises outlier noise points Finally, using algorithm normal angle threshold set 5.0/180.0* M_PI curvature 1.5 helps avoid under-segmentation over-segmentation issues, achieving complete leaves, LCCP fully segments proposed provides insights improve subsequent extraction, such as area, beneficial reconstruction rapeseed.

Language: Английский

---

A Hybrid Method for Individual Tree Detection in Broadleaf Forests Based on UAV-LiDAR Data and Multistage 3D Structure Analysis

Forests, Journal Year: 2024, Volume and Issue: 15(6), P. 1043 - 1043

Published: June 17, 2024

Individual tree detection and segmentation in broadleaf forests have always been great challenges due to the overlapping crowns, irregular crown shapes, multiple peaks large crowns. Unmanned aerial vehicle (UAV)-borne light ranging (LiDAR) is a powerful tool for acquiring high-density point clouds that can be used both trunk segmentation. A hybrid method combines proposed detect individual trees based on UAV-LiDAR data. distribution indicator-based approach first applied potential positions. The treetops extracted from canopy height model (CHM) segments obtained by applying marker-controlled watershed CHM are identify potentially false Finally, three-dimensional structures of trunks branches analyzed at each position distinguish between true was evaluated three plots subtropical urban with varying proportions evergreen trees. F-score ranged 0.723 0.829, which higher values than F-scores derived treetop (0.518–0.588) cloud-based (0.479–0.514). influences resolution (0.25 0.1 m) data acquisition season (leaf-off leaf-on) final result were also evaluated. results indicated using 0.25 m resulted under-segmentation crowns F-scores. had small influence when method. needs specify parameters prior knowledge forest. In addition, small-scale forests. Further research should evaluate natural over areas, differ forest compared

Language: Английский

---

PRISMA Review: Drones and AI in Inventory Creation of Signage

Drones, Journal Year: 2025, Volume and Issue: 9(3), P. 221 - 221

Published: March 19, 2025

This systematic review explores the integration of unmanned aerial vehicles (UAVs) and artificial intelligence (AI) in automating road signage inventory creation, employing preferred reporting items for reviews meta-analyses (PRISMA) methodology to analyze recent advancements. The study evaluates cutting-edge technologies, including UAVs equipped with deep learning algorithms advanced sensors like light detection ranging (LiDAR) multispectral cameras, highlighting their roles enhancing traffic sign classification. Key challenges include detecting minor or partially obscured signs adapting diverse environmental conditions. findings reveal significant progress automation, notable improvements accuracy, efficiency, real-time processing capabilities. However, limitations such as computational demands variability persist. By providing a comprehensive synthesis current methodologies performance metrics, this establishes robust foundation future research advance automated infrastructure management improve safety operational efficiency urban rural settings.

Language: Английский

---

Tree Crown Prediction of Spruce Tree Using a Machine-Learning-Based Hybrid Image Processing Method

Urban forestry & urban greening, Journal Year: 2025, Volume and Issue: unknown, P. 128815 - 128815

Published: April 1, 2025

Language: Английский

---

Tree Species Identification Using UAV RGB and Multispectral Cloud Points

Lecture notes in networks and systems, Journal Year: 2025, Volume and Issue: unknown, P. 66 - 79

Published: Jan. 1, 2025

Language: Английский

---

TreeLearn: A deep learning method for segmenting individual trees from ground-based LiDAR forest point clouds

Ecological Informatics, Journal Year: 2024, Volume and Issue: unknown, P. 102888 - 102888

Published: Nov. 1, 2024

Language: Английский

---