基于Mask RCNN和视觉技术的玉米种子发芽自动检测方法

doi:10.3969/j.issn.1004-1524.20221222

浙江农业学报 ›› 2023, Vol. 35 ›› Issue (8): 1927-1936.DOI: 10.3969/j.issn.1004-1524.20221222

基于Mask RCNN和视觉技术的玉米种子发芽自动检测方法

马启良^a(), 杨小明^a, 胡水星^a, 黄子鸿^b, 祁亨年^b^,^c^,^*()

湖州师范学院 a. 信息技术中心;b. 信息工程学院;c. 研究生院,浙江湖州 313000

收稿日期:2022-08-22 出版日期:2023-08-25 发布日期:2023-08-29
作者简介:马启良(1986—),男,安徽六安人,硕士,讲师,主要从事农业遥感与自动化研究。E-mail:mql@zjhu.edu.cn
通讯作者: *祁亨年,E-mail:02466@zjhu.edu.cn
基金资助:
浙江省重点研发计划(2019C02013);湖州市自然科学资金(2021YZ18)

Automatic detection method of corn seed germination based on Mask RCNN and vision technology

MA Qiliang^a(), YANG Xiaoming^a, HU Shuixing^a, HUANG Zihong^b, QI Hengnian^b^,^c^,^*()

a. Information Technology Center; b. School of Information Engineering; c. Postgraduate School, Huzhou University, Huzhou 313000, Zhejiang, China

Received:2022-08-22 Online:2023-08-25 Published:2023-08-29

摘要/Abstract

摘要：

种子标准发芽试验中,为获取种子发芽和生长情况,需借助人工定时对种子的发芽率、发芽势、芽长和根长等相关指标进行统计和测量,该测定过程费时费力,且易对发芽的幼苗造成损伤。针对这些问题,该研究基于Mask RCNN(基于区域的卷积神经网络)模型和机器视觉技术设计了一种玉米种子发芽自动检测方法。首先,在玉米种子发芽试验的7 d内,每天采集模型训练和测试所需的图像,并用Labelme工具对种子位置进行标注,再利用标注图像训练种子定位模型;其次,根据模型定位出的玉米种子掩膜区域,设定一个监测种子发芽的椭圆区域,自动识别种子发芽状态;最后,利用骨架提取和深度搜索算法实现发芽种子幼苗主骨架线的提取,通过计算种子掩膜的质心坐标位置,实现芽和根长度的分别统计。结果表明,该方法能够有效识别发芽种子,实现发芽试验中玉米种子的发芽率、发芽势、芽长、根长等指标的自动统计,可为种子发芽试验的自动化管理提供技术参考。

关键词: 标准发芽试验, 发芽率, Mask RCNN, 玉米, 骨架提取, 芽长, 根长

Abstract:

In the standard germination test of seed, manual timing is required to collect data on seed germination and growth, such as germination rate, germination vigor, shoot length, and root length. However, this measurement process is time-consuming, laborious, and can easily damage germinating seedlings. To address these issues, a automatic detection method for maize seed germination is designed based on the Mask RCNN (region-based convolutional neural network) model and machine vision technology. First, within the 7-day germination test period of maize seeds, images for model training and testing are collected daily, and the seed positions are annotated using the Labelme tool, then a seed localization model is trained based on the annotated images. Second, based on the seed mask regions located by the model, an elliptical region for monitoring seed germination is defined, and the seed germination status is automatically identified. Finally, the main skeleton line of germinating seedlings is extracted using skeleton extraction and depth-first search algorithms, shoot length and root length are measured separately by calculating the centroid coordinates of seed masks. The results show that this method can effectively recognize germinating seeds and automatically measure indicators such as germination rate, germination vigor, shoot length and root length in maize seed germination experiments, providing a technical reference for the automation management of seed germination tests.

Key words: standard germination test, germination rate, mask region-based convolutional neural network, corn, skeleton extraction, shoot length, root length

中图分类号:

S513
S237

马启良, 杨小明, 胡水星, 黄子鸿, 祁亨年. 基于Mask RCNN和视觉技术的玉米种子发芽自动检测方法[J]. 浙江农业学报, 2023, 35(8): 1927-1936.

MA Qiliang, YANG Xiaoming, HU Shuixing, HUANG Zihong, QI Hengnian. Automatic detection method of corn seed germination based on Mask RCNN and vision technology[J]. Acta Agriculturae Zhejiangensis, 2023, 35(8): 1927-1936.

图/表 11

图1 基于Mask RCNN的种子发芽自动检测流程

Fig.1 Automatic detection process of seed germination based on Mask RCNN

图2 损失收敛与PR曲线 a,训练损失收敛曲线;b,测试集PR曲线。

Fig.2 Loss convergence and PR curve a, Training loss convergence curve; b, Test set precision recall curve.

图3 孔洞填充过程图像 a,二值图像;b,漫水算法处理后孔洞图像;c,合并后图像。

Fig.3 The process images of hole filling a, Binary image; b, Hole image processed by flooding algorithm; c, Merged image.

图4 种子定位与发芽种子图像分割结果从左到右依次为种子定位、种子芽根二值图像,以及发芽种子芽、根分割图像。

Fig.4 Results of seed positioning and germinated seeds image segmentation From left to right, it is seed positioning, seed bud root binary image, and germinated seed bud and root segmentation image.

图5 半径绝对差值、质心距离与标注掩膜外接圆半径比值的结果统计

Fig.5 Result statistics of the ratio of the absolute difference of radius and the centroid distance to the circumscribed circle radius of the marked mask

图6 种子发芽状态自动识别结果红色椭圆轮廓表示种子未发芽,绿色椭圆轮廓表示种子已发芽。

Fig.6 Automatic identification results of seed germination status The red elliptical contour indicates that the seed has not germinated, and the green elliptical contour indicates that the seed has germinated.

表1 人工统计和算法自动统计的发芽种子数量与占比

Table 1 Germinated seed numbers and ratios by manual and algorithm automatic statistics

方法 Method	第2天Day 2		第3天Day 3		第4天Day 4		第5天Day 5		第6天Day 6		第7天Day 7
方法 Method	数量 Number	占比 Ratio/%	数量 Number	占比 Ratio/%	数量 Number	占比 Ratio/%	数量 Number	占比 Ratio/%	数量 Number	占比 Ratio/%	数量 Number	占比 Ratio/%
人工统计	17	34	48	96	50	100	50	100	50	100	50	100
Manual statistics
算法统计	8	16	37	74	47	94	50	100	50	100	50	100
Algorithm statistics

图7 人工和算法自动统计种子发芽数量结果比较

Fig.7 Result comparison of germinated seed numbers between manual and algorithm automatic statistics

图8 玉米种子识别定位、芽根图像的分割和骨架提取结果

Fig.8 Corn seed recognition positioning, bud root image segmentation and skeleton extraction results

图9 人工测量和算法测量的芽长、根长比较

Fig.9 Comparison of bud length and root length measured by manual measurement and algorithm measurement

表2 玉米幼苗芽长、根长误差分析

Table 2 Error analysis of bud length and root length of corn seedling

编号 Number	最大绝对误差 Maximum absolute error/mm		最小绝对误差 Minimum absolute error/mm
编号 Number	芽Bud	根Root	芽Bud	根Root
1~5	1.30	4.42	0.26	2.6
6~10	1.56	4.55	0.52	1.30
10~15	1.05	4.29	0.28	1.97
16~20	1.38	4.17	0.16	2.05

参考文献 21

[1]	贾良权, 祁亨年, 胡文军, 等. 采用TDLAS技术的玉米种子活力快速无损分级检测[J]. 中国激光, 2019, 46(9): 297-305.
	JIA L Q, QI H N, HU W J, et al. Rapid nondestructive grading detection of maize seed vigor using TDLAS technique[J]. Chinese Journal of Lasers, 2019, 46(9): 297-305. (in Chinese with English abstract)
[2]	李美凌, 邓飞, 刘颖, 等. 基于高光谱图像的水稻种子活力检测技术研究[J]. 浙江农业学报, 2015, 27(1): 1-6.
	LI M L, DENG F, LIU Y, et al. Study on detection technology of rice seed vigor based on hyperspectral image[J]. Acta Agriculturae Zhejiangensis, 2015, 27(1): 1-6. (in Chinese with English abstract)
[3]	王冬, 王坤, 吴静珠, 等. 基于光谱及成像技术的种子品质无损速测研究进展[J]. 光谱学与光谱分析, 2021, 41(1): 52-59.
	WANG D, WANG K, WU J Z, et al. Progress in research on rapid and non-destructive detection of seed quality based on spectroscopy and imaging technology[J]. Spectroscopy and Spectral Analysis, 2021, 41(1): 52-59. (in Chinese with English abstract)
[4]	孙俊, 张林, 周鑫, 等. 采用高光谱图像深度特征检测水稻种子活力等级[J]. 农业工程学报, 2021, 37(14): 171-178.
	SUN J, ZHANG L, ZHOU X, et al. Detection of rice seed vigor level by using deep feature of hyperspectral images[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(14): 171-178. (in Chinese with English abstract)
[5]	傅丹桂, 孙雁, 黄正仙, 等. 水稻种子不同活力测定方法的比较[J]. 云南农业大学学报(自然科学), 2018, 33(5): 811-817.
	FU D G, SUN Y, HUANG Z X, et al. Comparative research on the testing methods of different seed vigor in rice[J]. Journal of Yunnan Agricultural University(Natural Science), 2018, 33(5): 811-817. (in Chinese with English abstract)
[6]	徐振飞, 过晟鹏, 钱旺, 等. 3种杉木种子活力测定方法比较[J]. 浙江农林大学学报, 2020, 37(6): 1230-1234.
	XU Z F, GUO S P, QIAN W, et al. A comparative study of three determination methods for seed vigor of Cunninghamia lanceolata[J]. Journal of Zhejiang A & F University, 2020, 37(6): 1230-1234. (in Chinese with English abstract)
[7]	高嘉轩. 基于机器视觉的类圆形水果缺陷检测关键技术的研发与实现[D]. 南京: 南京邮电大学, 2020.
	GAO J X. Research and implementation of key techniques for defect detection of round-like fruits based on machine vision[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2020. (in Chinese with English abstract)
[8]	张晗, 王成, 董宏图, 等. 基于机器视觉的白菜种子精选方法研究[J]. 农机化研究, 2021, 43(12): 31-36.
	ZHANG H, WANG C, DONG H T, et al. Study on the seed selection method of cabbage based on machine vision[J]. Journal of Agricultural Mechanization Research, 2021, 43(12): 31-36. (in Chinese with English abstract)
[9]	潘霞, 谭会君. 计算机视觉技术在玉米种子自动检测中的应用[J]. 农机化研究, 2019, 41(3): 228-231.
	PAN X, TAN H J. Application of computer vision technology in maize seed automatic detection[J]. Journal of Agricultural Mechanization Research, 2019, 41(3): 228-231. (in Chinese with English abstract)
[10]	杨丽丽, 张大卫, 罗君, 等. 基于SVM和AdaBoost的棉叶螨危害等级识别[J]. 农业机械学报, 2019, 50(2): 14-20.
	YANG L L, ZHANG D W, LUO J, et al. Automatic recognition for cotton spider mites damage level based on SVM and AdaBoost[J]. Transactions of the Chinese Society for Agricultural Machinery, 2019, 50(2): 14-20. (in Chinese with English abstract)
[11]	张万红. 基于图像法的离体小麦叶片几何参数计算[J]. 浙江大学学报(农业与生命科学版), 2018, 44(6): 748-754.
	ZHANG W H. Geometrical parameter calculation of excised wheat leaves based on image analysis[J]. Journal of Zhejiang University(Agriculture and Life Sciences), 2018, 44(6): 748-754. (in Chinese with English abstract)
[12]	金沙沙, 贾良权, 龙伟, 等. 基于特征选择与骨架提取的种子萌发的芽长、根长检测[J]. 江苏农业学报, 2021, 37(3): 597-605.
	JIN S S, JIA L Q, LONG W, et al. Detection of seed bud length and root length based on feature selection and skeleton extraction[J]. Jiangsu Journal of Agricultural Sciences, 2021, 37(3): 597-605. (in Chinese with English abstract)
[13]	马启良, 胡水星, 林冬茂, 等. CSA-FFCM算法在玉米种子芽根长度自动化测定中的应用[J]. 湖州师范学院学报, 2021, 43(4): 42-49.
	MA Q L, HU S X, LIN D M, et al. Application of CSA FFCM algorithm in the automatic determination of corn seed bud root length[J]. Journal of Huzhou University, 2021, 43(4): 42-49. (in Chinese with English abstract)
[14]	吴旭东, 张晗, 罗斌, 等. 基于机器视觉的小麦种子活力检测方法[J]. 江苏农业科学, 2021, 49(24): 189-194.
	WU X D, ZHANG H, LUO B, et al. Study on wheat seed vigor detection method based on machine vision[J]. Jiangsu Agricultural Sciences, 2021, 49(24): 189-194. (in Chinese)
[15]	范丽丽, 赵宏伟, 赵浩宇, 等. 基于深度卷积神经网络的目标检测研究综述[J]. 光学精密工程, 2020, 28(5): 1152-1164.
	FAN L L, ZHAO H W, ZHAO H Y, et al. Survey of target detection based on deep convolutional neural networks[J]. Optics and Precision Engineering, 2020, 28(5): 1152-1164. (in Chinese with English abstract)
[16]	HE K M, GKIOXARI G, DOLLÁR P, et al. Mask R-CNN[C]// 2017 IEEE International Conference on Computer Vision (ICCV). October 22-29, 2017, Venice, Italy. IEEE, 2017: 2980-2988.
[17]	杜文圣, 王春颖, 朱衍俊, 等. 采用改进Mask R-CNN算法定位鲜食葡萄疏花夹持点[J]. 农业工程学报, 2022, 38(1): 169-177.
	DU W S, WANG C Y, ZHU Y J, et al. Fruit stem clamping points location for table grape thinning using improved mask R-CNN[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(1): 169-177. (in Chinese with English abstract)
[18]	冯青春, 成伟, 李亚军, 等. 基于Mask R-CNN的番茄植株整枝操作点定位方法[J]. 农业工程学报, 2022, 38(3): 128-135.
	FENG Q C, CHENG W, LI Y J, et al. Method for identifying tomato plants pruning point using Mask R-CNN[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(3): 128-135. (in Chinese with English abstract)
[19]	孙建波, 张叶, 常旭岭. 基于改进Mask R-CNN+LaneNet的车载图像车辆压线检测[J]. 光学精密工程, 2022, 30(7): 854-868.
	SUN J B, ZHANG Y, CHANG X L. Vehicle pressure line detection based on improved Mask R-CNN+LaneNet[J]. Optics and Precision Engineering, 2022, 30(7): 854-868. (in Chinese with English abstract)
[20]	袁山, 汤浩, 郭亚. 基于改进Mask R-CNN模型的植物叶片分割方法[J]. 农业工程学报, 2022, 38(1): 212-220.
	YUAN S, TANG H, GUO Y. Segmentation method for plant leaves using an improved Mask R-CNN model[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(1): 212-220. (in Chinese with English abstract)
[21]	李振, 廖同庆, 冯青春, 等. 基于机器视觉的蔬菜种子活力指数检测算法研究及系统实现[J]. 浙江农业学报, 2015, 27(12): 2218-2224.
	LI Z, LIAO T Q, FENG Q C, et al. Study on vegetable seed vigor index detection algorithm and system realization based on machine vision[J]. Acta Agriculturae Zhejiangensis, 2015, 27(12): 2218-2224. (in Chinese with English abstract)

基于Mask RCNN和视觉技术的玉米种子发芽自动检测方法

Automatic detection method of corn seed germination based on Mask RCNN and vision technology

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 21

相关文章 15

编辑推荐

Metrics

本文评价

[1]	雷联. 膜下滴灌调亏对制种玉米植株生长、产量和水分利用的影响[J]. 浙江农业学报, 2023, 35(7): 1542-1549.
[2]	张淑红, 张运峰, 武秋颖, 高凤菊, 李亚子, 纪景欣, 许可, 范永山. 玉米大斑病菌醇脱氢酶基因家族的鉴定和生物信息学分析[J]. 浙江农业学报, 2023, 35(5): 1108-1115.
[3]	王宁柯, 张瑞, 章胜勇. 机械化服务程度和农地经营规模对玉米生产效率的影响[J]. 浙江农业学报, 2023, 35(3): 698-707.
[4]	郑冉, 吕丹, 武清贵, 邸晓红, 朱通通, 邱冠杰, 罗红兵. 玉米C型胞质不育系S37-2败育的生物学与生理生化机制分析[J]. 浙江农业学报, 2023, 35(2): 259-265.
[5]	李娅楠, 冶文兴, 朱相德, 陈林, 徐晓锋, 张力莉. 基于LC-MS/MS技术研究稻草替代部分玉米青贮对奶牛血浆代谢产物的影响[J]. 浙江农业学报, 2023, 35(2): 266-274.
[6]	孟繁昊, 杨恒山, 张瑞富, 张玉芹, 李维敏, 张雨珊, 张明伟. 灌溉方式对西辽河平原玉米产量和水氮利用效率的影响[J]. 浙江农业学报, 2022, 34(9): 1826-1836.
[7]	姜昊梁, 黄允, 梁绍芳, 谢梦晨, 徐天成, 宋芷婷, 向文文, 陈青春, 万小荣, 孙伟. 镉胁迫对不同甜玉米自交系幼苗生长的影响及其相关简单重复序列分子标记初筛[J]. 浙江农业学报, 2022, 34(8): 1582-1590.
[8]	任梦云, 杜龙岗, 王美兴, 黄益峰. 糯玉米可溶性糖组分特征与采后品质特性[J]. 浙江农业学报, 2022, 34(6): 1133-1140.
[9]	胡开博, 杨清夏, 李扬, 吴开贤, 赵平, 龙光强. 化肥减氮配施氨基酸肥料对春玉米生产的影响[J]. 浙江农业学报, 2022, 34(4): 661-670.
[10]	贾利强, 赵秋芳, 陈曙, 丁波. 玉米转录因子bZIP G亚家族基因的表达模式[J]. 浙江农业学报, 2022, 34(2): 221-231.
[11]	崔文芳, 陈静, 鲁富宽, 秦丽, 秦德志, 王利平, 高聚林. 生物炭结合氮肥减量对玉米产量和氮效率的影响[J]. 浙江农业学报, 2022, 34(2): 248-254.
[12]	袁崇渊, 祝愿飞, 陈霞, 朱婵, 王毅, 陶海燕, 余娇娇. 玉米ZmNLP5与ZmSTP1、ZmAAP2基因启动子区域相互作用的鉴定[J]. 浙江农业学报, 2022, 34(11): 2340-2347.
[13]	高鑫, 杨恒山, 张瑞富, 张玉芹, 李锐, 张明伟. 浅埋滴灌水肥优化高产模式下春玉米产量与根冠特征差异性[J]. 浙江农业学报, 2022, 34(1): 1-9.
[14]	单长林, 周圆, 任琰, 季文彬, 李孝军. 玉米内州萎蔫病菌荧光重组酶介导等温扩增检测方法的建立[J]. 浙江农业学报, 2021, 33(9): 1676-1685.
[15]	王佳, 慕瑞瑞, 杨乔乔, 刘伟, 张月荷, 康建宏. 滴灌水肥一体化下施钾量对宁夏春玉米叶绿素荧光特性与产量的影响[J]. 浙江农业学报, 2021, 33(8): 1347-1357.