摘要
芽长是种子活力的一个重要判定标准。传统的芽长检测方法采用人工测量方式,存在费时费力、且受人为主观因素影响较大等问题。利用数字图像处理技术的芽长自动检测算法可以提高芽长测量的效率,并且能够统一测量标准从而避免主观误差。本研究基于特征选择与骨架提取算法原理设计了种子芽长、根长检测复合算法及软件,首先利用颜色特征提取叶片信息,并在整株芽长二值图像中去除叶片区域信息,其次通过圆盘结构元素与线性结构元素腐蚀图像分割出种子图像,获得种子的中心,再对整株芽长二值图像进行图像增强和边缘轮廓处理,最后对芽长图像进行骨架提取与剪枝,依据像素点间的欧氏距离计算芽长与根长。通过对玉米、小麦、水稻的芽长和根长进行测量,结果显示,玉米、小麦与水稻芽长的百分误差分别为2.90%、2.05%、2.40%;根长的百分误差分别为1.90%、2.11%、2.02%。说明基于特征选择与骨架提取检测方法的复合算法能够实现对种子萌发的芽长与根长的高精度、快速检测。
Bud length is an important criterion for seed vigor judgement.Traditional detection methods for bud length used manual measurements,which had the problems such as time consuming and the measuring results were affected by experimenters’subjective factors greatly.Automatic detection algorithm of bud length using digital image processing technology could improve the detection efficiency and unify the measurement standard to avoid subjective error.A compound algorithm and software for seed bud length and root length detection were designed based on the principle of feature selection and skeleton extraction algorithm.Firstly,the leaf information was extracted based on the color feature,and the leaf region information was wiped off in the binary image of bud length for the whole plant.Secondly,image of the seed was segmented based on eroded images of disk structure element and linear structure element,and the center of the seed was got.Then the binary image for bud length of whole plant was enhancement and the edge contour of the image was processed.After that,skeleton of the bud length image was extracted and the image was pruned,the bud length and root length were calculated according to the Euclidean distance between pixels in the skeleton.Bud length and root length of corn,wheat and rice were measured in the study.The results showed that,average percentage error of the bud length of corn,wheat and rice were 2.90%,2.05%and 2.40%respectively,while for root length the data were 1.90%,2.11%and 2.02%respectively.The experimental results show that,the compound algorithm based on feature extraction and skeleton extraction can detect the bud length and root length during seed germination precisely and fastly.
作者
金沙沙
贾良权
龙伟
祁亨年
赵光武
高璐
蒋林华
JIN Sha-sha;JIA Liang-quan;LONG Wei;QI Heng-nian;ZHAO Guang-wu;GAO Lu;JIANG Lin-hua(School of Information Engineering,Huzhou University,Huzhou 313100,China;Zhejiang Provincial Key Laboratory of Smart Management&Application of Modern Agricultural Resources,Huzhou 313100,China;College of Agriculture and Food Sciences,Zhejiang A&F University,Hangzhou 311000,China)
出处
《江苏农业学报》
CSCD
北大核心
2021年第3期597-605,共9页
Jiangsu Journal of Agricultural Sciences
基金
国家自然科学基金青年基金项目(31701512)
浙江省重点研发项目(2019C02013、2020C02020)
湖州市公益性应用研究项目(2019GZ15)
浙江省教育厅一般科研项目(Y201941626)。
关键词
图像处理
芽长检测
根长检测
颜色特征
骨架提取
image processing
bud length
root length
color feature
skeleton extraction
