基于三维点云的苹果树冠层光照分布模型研究
发布时间：2018-10-18 19:17
【摘要】：为给果园精细管理中果树修枝整形、果实品质评价以及果实产量估算等提供科学的理论依据和技术指导,以果园自然开心形苹果树为研究对象,基于果树三维点云结构,进行果树冠层空间光照分布建模研究。用三维点云重构技术和点云分割技术获取果树不同高度的点云分层,分别使用像素占比和Graham扫描算法计算各高度点云分层垂直投影的有效投影面积和占地面积及有效叶面积指数。以果树冠层不同高度层的有效叶面积指数为自变量,对不同高度层平均相对光照强度进行线性回归,获得果树冠层光照分布模型,并对模型进行验证。结果表明:所建果树冠层光照分布模型的校正决定系数R2c为0.924,校正均方根误差RMSEC为0.05,验证决定系数R2v为0.955,验证均方根误差RMSEP为0.04,相对分析误差RPD为4.91。该模型具有较高的预测精度和较强的预测能力。
[Abstract]:In order to provide scientific theoretical basis and technical guidance for fruit tree pruning and shaping, fruit quality evaluation and fruit yield estimation in the fine management of orchard, the natural happy apple tree in orchard was taken as the research object, based on the three-dimension point cloud structure of fruit tree. The spatial light distribution modeling of fruit tree canopy was carried out. Three dimensional point cloud reconstruction technique and point cloud segmentation technique were used to obtain point cloud stratification of different height of fruit trees. The effective projection area, the occupied area and the effective leaf area index of each height point cloud stratified vertical projection are calculated by using pixel duty ratio and Graham scanning algorithm respectively. Taking the effective leaf area index of different height layers of fruit tree as independent variable, the average relative light intensity of different height layers was linear regressed, and the light distribution model of fruit tree canopy was obtained, and the model was verified. The results showed that the calibration decision coefficient R2c, root mean square error (RMSEC), validation decision coefficient (R2v), root mean square error (RMSEP) and relative analysis error (RPD) of the established model were 0.924, 0.05, 0.955, 0.04 and 4.91 respectively. The model has high prediction accuracy and strong prediction ability.
【作者单位】： 中国农业大学信息与电气工程学院/现代精细农业系统集成研究教育部重点试验室;
【基金】：国家自然科学基金项目(31371532)
【分类号】：S661.1;TP391.41
，
本文编号：2280117
[Abstract]:In order to provide scientific theoretical basis and technical guidance for fruit tree pruning and shaping, fruit quality evaluation and fruit yield estimation in the fine management of orchard, the natural happy apple tree in orchard was taken as the research object, based on the three-dimension point cloud structure of fruit tree. The spatial light distribution modeling of fruit tree canopy was carried out. Three dimensional point cloud reconstruction technique and point cloud segmentation technique were used to obtain point cloud stratification of different height of fruit trees. The effective projection area, the occupied area and the effective leaf area index of each height point cloud stratified vertical projection are calculated by using pixel duty ratio and Graham scanning algorithm respectively. Taking the effective leaf area index of different height layers of fruit tree as independent variable, the average relative light intensity of different height layers was linear regressed, and the light distribution model of fruit tree canopy was obtained, and the model was verified. The results showed that the calibration decision coefficient R2c, root mean square error (RMSEC), validation decision coefficient (R2v), root mean square error (RMSEP) and relative analysis error (RPD) of the established model were 0.924, 0.05, 0.955, 0.04 and 4.91 respectively. The model has high prediction accuracy and strong prediction ability.
【作者单位】： 中国农业大学信息与电气工程学院/现代精细农业系统集成研究教育部重点试验室;
【基金】：国家自然科学基金项目(31371532)
【分类号】：S661.1;TP391.41
，
本文编号：2280117
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2280117.html
