自然环境下苹果点云多维度特征分割方法研究  

Research on apple point cloud segmentation techniques in natural environments

作　　者：李娜[1,2] 安楠 张立杰 姜海勇 陈广毅[1] 施宇 LI Na;AN Nan;ZHANG Lijie;JIANG Haiyong;CHEN Guangyi;SHI Yu(College of Mechanical and Electrical Engineering,Hebei Agricultural University,Baoding 071001,China;Hebei Innovation Center of Smart Agricultural Equipment Technology,Baoding 071001,China)

机构地区：[1]河北农业大学机电工程学院,河北保定071001 [2]河北省智慧农业装备技术创新中心,河北保定071001

出　　处：《河北农业大学学报》2024年第3期105-112,共8页Journal of Hebei Agricultural University

基　　金：河北省重点研发计划项目(21321902D).

摘　　要：为了解决自然环境下苹果果园复杂场景及光照变化对果实精准定位和空间形态评估带来的困难,对基于多维度特征的苹果点云分割方法进行了研究。研究中,通过融合欧氏距离、曲率分析和颜色特征创建苹果点云分割掩膜,对苹果点云进行分割;引入K-D Tree进行聚类修正,拟合后获取最终果实空间全面信息。试验结果显示:在自然果园环境下,该方法在逆光、顺光和侧光条件下分别取得了96.20%、97.67%和97.93%的分割纯净率,与仅基于欧氏距离或颜色特征的单一特征分割方法相比,该方法的纯净率分别提高了37.57%和14.53%,且聚类误分问题得到有效解决。该方法具有良好的鲁棒性和精确性,可为苹果智能化采摘作业的精确性和可靠性提供技术支持。Due to the complexity of apple orchard scenes in natural environments and factors such as changes in illumination,there are challenges in precise positioning and spatial morphological evaluation of fruits for intelligent picking operations.This article proposed a method of apple point cloud segmentation based on multi-dimensional features.This method creates apple point cloud segmentation masks by integrating Euclidean distance,curvature analysis,and color features to segment the apple point cloud.It also introduced K-D Tree for clustering correction to obtain comprehensive spatial information of the final fruit after fitting.Experimental results showed that the segmentation purity rates of this method were 96.20%,97.67%,and 97.93%under backlight,frontlight and sidelight conditions in natural orchard,Compared with the single feature segmentation method based only on Euclidean distance or color features,the purity rate of this method was improved by 37.57%and 14.53%,respectively.And the problem of clustering misclassification has been effectively solved.This algorithm displayed good robustness and accuracy,which can provide technical support for the precision and reliability of apple intelligent picking operations.

关 键 词：双目相机 点云分割 识别定位 特征融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]