摘要
针对图像法根系表型检测中因土壤遮挡而导致根系图像断裂的问题,提出一种基于腐蚀生长算法的玉米根系修复方法,并进行了不同活力玉米种子早期根系表型研究。首先,采用长方形扁平结构透明培养容器种植玉米,迫使其根系贴壁生长,可得到清晰的玉米根系图;通过偏振镜和单反相机采集图像,并采用灰度化、二值化、水漫算法等对图像进行预处理,可有效去除因设备反光和土壤色差造成的各类噪声。其次,基于玉米根系的向水性、向地性、连续性等生理特性,提出5条图像修补规则,即端点判定规则、分叉点判定规则、内部连续性规则、片段生长规则、近邻生长规则,在以上规则约束下,通过细化图像得到单像素连接的根系骨干,以各个根段的末端点为起点向中心腐蚀,并为属于不同根段的点集标记不同编号,根据不同根段间的端点导数值和平均导数值等参数,连接根段,实现根系的修补,从而得到完整根系图像。最后,基于所提图像修复算法对不同活力的玉米种子根系图像进行表型研究,发现在相同时间,根系数目、根系宽度、根系长度、根系延展长度与玉米种子活力均呈现明显负相关;以上4个生理参数的增速与种子活力呈现明显正相关。研究表明,本文所提的根系修复算法可用于作物根系高通量表型无损检测。
The root phenotypes of different vigorous maize seeds vary a lot,and imaging roots of growing maize is a non-invasive,affordable and high throughput way to detect it. However,due to the block of soil,acquiring a complete image is difficult. An algorithm was proposed to repair incomplete root images and based on it, root fast non-invasive phenotyping detection can be realized. Firstly, a cuboid transparent container without cover was developed as mesocosms and the maize seeds were planted in it.The maize roots grew in soil against two acrylic plastic surfaces due to the press of the small growing area to acquire more root details during roots visualization and imaging. Even though,parts of the roots were occluded by the soil which meant that it was tough to extract the information of root general physical construction. For recovering gaps from disconnected root segments,corrosion growth algorithm was proposed based on the physiological characteristics of hydrotropism,geostrophic and continuity with three steps which were root image thinning,corrosion and growing processing,respectively. The experiments indicated that maize phenotyping parameters were negatively correlated with seed aging days. And specifically,root number,root length,root width and root extension length of unaged and 14-day-aged maize seeds were decreased from 14. 80,83. 50 mm,1. 53 mm and 82. 70 mm to 4. 38,36. 90 mm,1. 38 mmand 54. 6 mm,and the growing speed of them were changed from 1. 68 per day,8. 80 mm/d,0. 06 mm/d,9. 0 mm/d to 0. 70 per day,4. 3 mm/d,0. 05 mm/d and 5. 70 mm/d,respectively.Whereas root extension angle is basically irrelevant with the level of maize seed aging. The developed cuboid transparent container without cover can push each root growing along the wall of the container which helped to acquire more root information. The presented novel corrosion growth algorithm can recover the missing parts,even for big gaps,of maize roots effectively according to root morphological properties. The experiments showed that the proposed method can be applied to evaluate the vigor of maize seeds which had vast application prospect in high throughput root phenotyping area.
作者
卢伟
李也
王玲
罗慧
孙国祥
汪小旵
LU Wei;LI Ye;WANG Ling;LUO Hui;SUN Guoxiang;WANG Xiaochan(College of Engineering,Nanjing Agricultural University,Nanjing 210031,China;Engineering Laboratory of Modern Facility Agriculture Technology and Equipment in Jiangsu Province,Nanjing 210031,China)
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2020年第4期224-231,共8页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金项目(11604154)
江苏省自然科学基金面上项目(BK20181315、BK20170727)
江苏省农机三新工程项目(SZ120170036)
Asia Hub南京农业大学-密歇根州立大学联合研究项目(2017-AH-11)
扬州市重点研发计划(现代农业)项目(YZ2018038)。
关键词
玉米种子
根系表型
种子活力
腐蚀生长算法
图像修复
maize seed
root phenotype
seed vigor
corrosion growth algorithm
image repairing