基于三维点云颜色特征的苹果树冠层光照分布计算方法

Calculation of Light Distribution of Apple Tree Canopy Based on Color Characteristics of 3D Point Cloud

摘要:

合理的果树冠层结构有利于获取充足的光照，对提升果实产量及品质具有重要意义。为了揭示果树冠层光照分布规律，以自由纺锤形苹果树为研究对象，以目标图像的颜色变化与光照强弱存在相关性为理论依据，首先利用Trimble TX5型地面三维激光扫描仪以“顶视法”获取叶幕稳定期苹果树冠层三维点云，按照实际苹果树冠层划分方法，提取三维点云空间不同区域的颜色特征，针对自然环境下苹果树冠层颜色特征具有复杂性和模糊性，不能采用精确、定量的符号对其进行描述的不足，构建以颜色特征为输入、相对光照强度为输出的模糊神经网络，以此作为苹果树冠层光照分布预测模型。试验结果表明：提出的基于三维点云颜色特征的光照分布计算方法具有较好的可行性，预测精度为80.57%，能够为科学合理的苹果树修剪和整形提供技术支撑。

Abstract:

Study on light distribution of apple tree canopy is one of the important ways to optimize type of fruit trees and improve potential production. The reasonable and effective use of light energy and optimization of light distribution within fruit tree canopy have vital significance for the formation of fruit tree growth and fruit quality. Calculation method of light distribution of free spindle apple canopy was carried out. In recent years, there were several researches on light distribution of plant canopy on the basis of three dimensional morphology of plant canopy by using mathematical simulation methods which could not express the real light distribution. In order to reveal real light distribution rule of canopy intelligently and efficiently, apple trees of spindle shape were selected as research objects, and based on the correlation that the target image color changed with the light intensity. Firstly, 3D point cloud of apple tree canopy in leaf curtain stability period was captured by Trimble TX5 laser scanner; secondly, according to the actual canopy division method, color information of different areas in 3D canopy space was extracted; thirdly, aiming at the shortcomings of complexity, fuzziness and indescribability by precise and quantitative symbols under natural environment, the fuzzy neural network was constructed to predict light distribution with color characteristic as input and relative light intensity as output. The experimental results showed that the precision of the proposed method had good feasibility, the prediction accuracy was 80.57%. The result of this study will provide theoretical basis for scientific pruning to get the best light distribution.