关键词：
簸箕柳;形态表型参数;图像处理;三维点云;无损测量
摘 要：
传统人工测量仪器提取植物形态表型参数的方式存在效率低、误差大、适用性弱、破坏性强等缺陷,成为制约植物表型研究发展的瓶颈。将模式树种杨树的姊妹种簸箕柳(Salix suchowensis Cheng)作为研究对象,设计了一套基于图像处理技术的植物形态表型参数获取系统,通过高分辨率RGB相机获取植株的二维图像序列,基于HSV模型阈值分割方法实现簸箕柳与背景环境的分割,结合运动中恢复结构算法生成三维点云,并利用棋盘格进行坐标系间的距离转换。使用本系统提取簸箕柳的株高、基径、叶片面积、分枝数和分枝角等表型参数并与传统人工接触式测量值相比较,平均绝对百分比误差分别为7.69%,13.94%,9.99%,5.32%,9.30%。结果表明,本研究设计的植物形态表型参数获取系统能够较好地重建植物三维点云,并可快速、准确提取植物的形态表型参数,能够实现无损测量,满足研究对象簸箕柳三维形态表型参数的测量和生长观测的需求,可为植物尤其是林木表型研究提供数据参考。
译 名：
A method for obtaining plant morphological phenotypic parameters using image processing technology
作 者：
LI Yangxian;ZHANG Huichun;YANG Yang;College of Mechanical and Electronic Engineering,Nanjing Forestry University;
关键词：
Salix suchowensis Cheng;;morphological phenotypic parameter;;image processing;;3D point cloud;;nondestructive measurement
摘 要：
China is a large agriculture and forestry country.Agroforestry is the foundation of China's national economy.However,with the continuous change of the worldwide environment,the global temperature and carbon dioxide concentration are rising,the degree of land salinization is increasing,and plants will face more and more harsh living environment.By studying the phenotype of plants,it can be understood the physical,physiological and biochemical characteristics and traits of plants produced by the interaction of the genotype of plants with the environment.Accelerating the study of plant phenotype can effectively promote the improvement of plant varieties,make plants adapt to the change of global environment,and thus accelerate the development of agriculture and forestry in China.The traditional method of extracting plant morphological phenotypic parameters by manual measuring instruments has many drawbacks,such as low efficiency,large error,weak applicability and strong destructiveness,which has become the bottleneck of plant phenotypic research.In this paper,Salix suchowensis Cheng,a sister species of the model tree poplar,was designed as a research object,and a set of plant morphological phenotypic parameters acquisition system based on the image processing technology was designed.The two-dimensional image sequence of the plant was obtained by the high-resolution RGB camera,the threshold segmentation method based on HSV model was used to segment the S.suchowensis Cheng and the background environment,and the three-dimensional point cloud was generated by combining the structure from the motion(SFM) algorithm,and used the checkerboard to perform distance conversion between coordinate systems.Parameters such as plant height,basal diameter,leaf area,branch number and branch angle phenotype of S.suchowensis Cheng were extracted by this system and compared with the traditional manual contact measurement values.The average absolute percentage errors were 7.69%,13.94%,9.99%,5.32% and 9.30%,respectively.The results showed that the plant morphological parameter acquisition system could reconstruct the three-dimensional point cloud of plant,and had the ability to extract the morphological phenotypic parameters of plant quickly and precisely.Meanwhile,the system can achieve non-destructive measurement and satisfy the demand for the measurement of the three-dimensional phenotypic parameters of S.suchowensis Cheng and the demand for S.suchowensis Cheng growth observation.Therefore,the system can provide data reference for the research on plant phenotype especially forest trees.
