基于随机森林回归算法的苹果树冠层光照分布模型  

Illumination Distribution Model of Apple Tree Canopy Based on Random Forest Regression Algorithm

作　　者：师翊 耿楠 胡少军[2,3] 张志毅 张晶[4] SHI Yi;GENG Nan;HU Shaojun;ZHANG Zhiyi;ZHANG Jing(College of Mechanical and Electronic Engineering,Northwest A&F University,Yangling,Shaanxi 712100,China;Key Laboratory of Agricultural Internet of Things,Ministry of Agriculture and Rural Affairs,Yangling,Shaanxi 712100,China;College of Information Engineering,Northwest A&F University,Yangling,Shaanxi 712100,China;Laboratory of Computer Science,Information Processing and Systems,National Institute of Applied Sciences of Rouen,Rouen 76800,France)

机构地区：[1]西北农林科技大学机械与电子工程学院,陕西杨凌712100 [2]农业农村部农业物联网重点实验室,陕西杨凌712100 [3]西北农林科技大学信息工程学院,陕西杨凌712100 [4]国立鲁昂应用科学学院信息与系统特征实验室,鲁昂76800

出　　处：《农业机械学报》2019年第5期214-222,共9页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金项目(61303124);中央高校基本科研业务费专项资金项目(Z109021708)

摘　　要：合理的果树冠层结构有利于光照的有效分布,对提升果实产量与品质有重要意义。为揭示果树冠层内部的光照分布情况,针对目前果树冠层内部光照强度获取难度大、预测精度低的问题,研究了冠层颜色特征与光照强度的对应关系,提出一种基于冠层剖面阴影特征和冠层点云颜色特征的随机森林预测模型。以纺锤形"陕富6号"苹果树为研究对象,首先使用Kinect 2. 0采集果树的双面点云数据,预处理后得到完整的点云数据;其次,基于改进的空间殖民算法和叶序添加规则重构果树的三维模型;最后,使用"切片法",在垂直方向上将冠层模型每0. 1 m分层划分,使用POV-Ray渲染器逐层渲染阴影,同时使用光照度计,自顶向下每0. 1 m实测光照强度数据,构建以每层阴影图灰度特征和每层点云HSI颜色特征为输入,以相对光照强度为输出的随机森林网络。试验结果表明,该方法能够较为准确地预测冠层内的光照分布情况,预测值与实际值的决定系数R^2为0. 864,平均绝对百分比误差MAPE为0. 236,RF回归模型可作为苹果树冠层内光照分布预测的有效方法,为果树的剪枝、整形等研究提供参考。The reasonable structure of fruit tree canopy is beneficial to the effective distribution of illumination, which has vital significance to enhance the fruit yield and quality. At present, it is difficult to obtain illumination intensity data in the canopy of fruit trees, and the prediction accuracy is low. In order to study inner canopy illumination distribution, a random forest prediction model was proposed based on canopy profile shadow feature and point cloud color feature. The detailed research methods were shown as follows. Firstly, the spindle “Shanfu 6” apple tree was chosen as the research object and Kinect 2.0 was used to acquire double face point cloud data of tree, and then the complete data was obtained with pre-process. Secondly, the improved space colonization algorithm with growth angle constraint and phyllotaxis adding rules were used to rebuild apple tree 3D model. Finally, the “slice method” was used to cut canopy model every 0.1 m in the vertical direction, and then the POV Ray renderer was used to render shadows layer after layer, meanwhile, light meter was used to obtain illumination intensity data every 0.1 m from top to bottom consistently, and the random forest network that with input data of color feature of every layer and output data of relative illumination intensity was built as the apple tree canopy illumination distribution prediction model. The experiment results showed that the proposed method can predict the illumination distribution accurately. The determination coefficient R 2 between true value and predicted value was 0.864, and MAPE was 0.236. Random forest regression model can be used as an efficient method for prediction of canopy illumination distribution, and it can provide reference for fruit tree pruning and plastic research.

关 键 词：苹果树 点云 三维重构 冠层阴影 冠层光照 随机森林 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] S24[自动化与计算机技术—计算机科学与技术]