基于轮廓分割的草莓叶片三维建模

发布时间：2018-01-19 12:42

本文关键词： 图像采集图像分割农作物多源信息点云聚类三维建模　出处：《农业工程学报》2017年S1期 　论文类型：期刊论文

【摘要】：为精确构建原位草莓植株三维空间结构,以高架栽培模式生长环境下的草莓植株为研究对象,提出了一种基于多源图像轮廓分割的草莓植株结构形态三维重建算法。通过改进的多源图像融合算法,建立多源图像映射关系,融合预处理后的多源信息得到待分割强度图;计算待分割强度图矢量场卷积的局部中心,选出多目标的初始轮廓控制点,将参数的活动轮廓模型应用于待分割强度图像进行叶片的分割;采用标记的方法将分割轮廓映射至距离点云集,设计以单个叶片为单位的平面拟合选择机制,最终完成草莓三维模型的重建及显示。为验证该算法的有效性,将三维重建后的有效叶片数,平均单叶长度及叶片距离差作为评价指标,实验结果表明,有效叶片数正确率为85.6%,平均单叶长度模型正确计算率为88.4%,叶片距离差正确计算率为82.4%,研究结果可应用于原位草莓植株的空间位置测量,可为农业机器人局部视觉场景中植株空间结构的构建提供参考。
[Abstract]:In order to accurately construct the three-dimensional spatial structure of strawberry plants in situ, strawberry plants grown in elevated culture mode were taken as the research object. A 3D reconstruction algorithm of strawberry plant structure and morphology based on multi-source image contour segmentation is proposed. The multi-source image mapping relationship is established through an improved multi-source image fusion algorithm. The preprocessed multi-source information is fused to obtain the intensity map to be segmented. The local center of vector field convolution of intensity map is calculated and the initial contour control point of multi-object is selected. The active contour model of parameters is applied to segmenting the blade of intensity image to be segmented. The segmentation contour is mapped to the distance point cluster by the method of marking, and a plane fitting selection mechanism based on a single blade is designed, and the reconstruction and display of the 3D model of strawberry are completed finally. The validity of the algorithm is verified. The effective leaf number, the average leaf length and the leaf distance difference after 3D reconstruction were taken as the evaluation indexes. The experimental results showed that the correct rate of effective leaf number was 85.6%. The correct calculation rate of average single leaf length model was 88.4 and the correct calculation rate of leaf distance difference was 82.4. The results can be used to measure the spatial position of strawberry plants in situ. It can provide a reference for the construction of plant spatial structure in the local vision scene of agricultural robot.
【作者单位】：中国农业大学现代精细农业系统集成研究教育部重点实验室;
【基金】：国家高技术研究发展计划(863计划)(2013AA102406)
【分类号】：TP391.41
【正文快照】： 张雪,郭彩玲,宗泽,张伟洁,刘刚.基于轮廓分割的草莓叶片三维建模[J].农业工程学报,2017,33(增刊1):206-211.doi:10.11975/j.issn.1002-6819.2017.z1.031 http://www.tcsae.orgZhang Xue,Guo Cailing,Zong Ze,Zhang Weijie,Liu Gang.3D reconstruction of strawberry leaves ba

【参考文献】