摘要
在单株水稻表型测量研究中,为了实现绿叶面积和茎叶相关表型参数的准确计算提供技术保障,茎叶的分割是非常重要的一步。传统的人工测量方法费时费力,且主观性较强,而基于普通相机拍摄的彩色图像进行分割效果很差。本研究介绍了一种使用可见光-近红外高光谱成像系统自动区分单株盆栽水稻茎叶的方法。首先将各波长下的图像从原始二进制数据中提取出来,接着使用主成分分析所有波长下的图像,并提取出主要的主成分图像,再基于数字图像处理技术将茎叶区分开。实验结果表明,本系统以及文中所用方法对分蘖盛期的水稻茎叶有很好的分割效果,这为后续水稻茎叶表型性状高通量、数字化、无损准确提取提供了重要的技术保障,并进一步促进植物表型组学的发展。
In the study of phenomics,the segmentation of leaf and stem of individual rice plant is very important, which can be used to provide basis for the calculation of phenotypic parameters,such as green leaf area and biomass.Traditional methods are subjective,time-consuming,and labor-intensive.The segmentation with color image,acquired by CCD camera,has shown a poor result.This study introduced an automatic segmentation method of leaf and stem with a hyperspectral imaging system.First,the images of individual rice plant under different wavelength were extracted from original binary stream.Then the principal component analysis （PCA）was used to analyze all the images and extract main principal component images.At last,these main images were used to segment the leaf and stem with digital image processing.The result has shown that this hyperspectral imaging system and method that was used in this study has good segmentation outcome for the leaf and stem of individual rice plant on the tillering stage.This work provides a break-through for high-throughput,non-destructive,and accurate extracting the leaf and stem of rice,and promotes the devel-opment of the plant phenomics.
出处
《激光生物学报》
CAS
2015年第1期31-37,共7页
Acta Laser Biology Sinica
基金
国家高技术研究发展计划(863计划
2013AA102403)
国家自然科学基金资助项目(30921091
31200274)
新世纪优秀人才支持计划(NCET-10-0386)
关键词
高光谱成像
图像分割
主成分分析
hyperspectral imaging
image process
principal component analysis
作者简介
冯慧（1987-），女，汉族，湖北浠水人，博士研究生，主要从事水稻表型组学研究。（手机）15071283065；（电子邮箱）fenghuith2006@126．com
通讯作者：杨万能（1984-），男，汉族，湖北南漳人，副教授，主要从事植物表型组学研究。（手机）15871800820；（电子邮箱）ywn@mail．hzau．edu．cn
