果树三维重建与冠层结构参数检测研究进展

Research Progress on Three-dimensional Reconstruction of Fruit-tree Canopies and Detection of Canopy Structure Parameters

摘要:

果树冠层结构是影响果品产量与品质的核心因素，对其进行三维重建可为果树精细化管理与智慧果园建设提供重要数据支持。传统人工测量效率低、误差大，难以适配现代果园管理需求;而基于激光雷达、无人机及多传感器融合的三维监测系统，凭借其高精度、自动化及可重复性优势，已广泛用于冠层体积、叶墙面积等结构参数的智能检测。为此，本文聚焦 "三维重建支撑树冠层结构参数检测" 核心主题，遵循 "设备一方法一流程" 理主线，系统回顾相关研究进展，并讨论总结当前存在挑战及未来发展趋势。首先，分类阐述了空中、地面及固定 / 移动扫描等三维重建数据采集设备与载体平台的功能特征及关键性能;其次，对比分析了影像、点云及深度学习辅助等重建方法的具体优势、局限性及适配应用场景;进一步，针对果树冠层点云构建的 "配准一地面分离一单株分割任务匹配重建" 标准流程，明确了典型结构参数的解析步骤及误差来源，阐明了覆盖率等质控指标对参数检测精度、技术普适能力的关联影响;最终，归纳了复杂冠层观测盲区、多平台协同等关键研究难题，并提出了统一标准与特定任务双重驱动的果树冠层重建与监测发展方向。

Abstract:

Fruit-tree canopy structure is a key determinant of fruit yield and quality, and its threedimensional (3D) reconstruction can provide essential data support for precision orchard management and the development of smart orchards. Traditional manual measurements are inefficient and error-prone, and are increasingly inadequate for moder orchard management, whereas 3D monitoring systems based on LiDAR, unmanned aerial vehicles (UAVs) and multi-sensor fusion have been widely used for the intelligent detection of canopy volume, leaf-wall area and other structural parameters owing to their advantages in accuracy, automation and repeatability. To this end, this paper focuses on the theme of 3D reconstruction for detecting the structural parameters of fruit-tree canopies. By folloving the "equipmentmethods-workflow" framework, it systematically reviews recent research progress and discusses current challenges and future development trends. Firstly, the functional characteristics and key performance metrics of aerial, ground-based and fixed/ mobile scanning devices and platforms for 3D data acquisition were categorized and described. Secondly, the image-based, point-cloud-based and deep-learningassisted reconstruction methods were comparatively analyzed in terms of their specific advantages, limitations and suitable application scenarios. Furthermore, for the standardized workflow of fruit-tree canopy point-cloud processing (" registration-ground separation-single-tree segmentation-task-oriented reconstruction"), the processing steps and error sources involved in deriving typical canopy structural parameters were clarified, and coverage and other quality-control indicators were used to jointly assess parameter-detection accuracy and the methodological generalizability across orchards and platforms. Finally, key research challenges such as occlusion in complex canopies and muli-platform coordination were summarized, and future development directions for fruit-tree canopy reconstruction and monitoring that were jointly driven by unified standards and task-specific requirements were proposed. This review provided a systematic technical framework for 3D information acquisition and standardized detection of canopy structural parameters in smart orchards, and offered effective support for fine-scale operations such as variable-rate spraying and pruning optimization.