刘春鹏

语义分割技术能够实现果树冠层图像的像素级识别，是智慧果园信息获取的重要技术手段。结合果树冠层图像特点，本文对语义分割技术涉及的图像采集、预处理、特征提取及应用进行了梳理，构建了果树冠层图像语义分割技术应用框架，分析了当前应用中存在的问题，并对未来发展方向进行了展望。研究结果可为果树冠层智能识别及相关技术应用提供参考。

果树冠层；计算机视觉；语义分割；图像处理；智慧农业

[1] Li,Z.,Deng,X.,Lan,Y.,et al.Fruit tree canopy segmentation from UAV orthophoto maps based on a lightweight improved U-Net. Computers and Electronics in Agriculture,2024,217:108538. [2] Leng,H.,Wang,Z.,He,M.,et al.Effects of different ground segmentation methods on UAV-based canopy measurements.Frontiers i n Plant Science,2024,15:1393592. [3] Luo,Z.,Yang,W.,Yuan,Y.,et al.Semantic segmentation of agricultural images:A survey.Information Processing in Agriculture,2024, 11(2):172–186. [4] Minaee,S.,et al.Image Segmentation Using Deep Learning:A Survey.IEEE TPAMI,2022,44(7):3523–3545. [5] Wei,J.,et al.Single tree semantic segmentation from UAV images based on improved U-Net network.Drones,2025,9(4):237.

点击下载PDF