摘要
分蘖数是冬小麦产量评估的重要指标,分蘖性状的高通量获取对小麦育种研究及功能基因组鉴定具有重要意义。针对目前小麦分蘖数检测主要依靠人工普查、费时耗力、可重复性差等问题,提出一种基于机器视觉的抽穗期冬小麦分蘖数快速检测方法。首先,利用超绿算子与RGB色彩空间下恒定颜色分割,结合形态学操作,获得目标前景区域和边界;然后,使用小波变换Haar特征的线特征检测拓展模板,提取完整植株边缘信息;最后,采用Hough直线检测算法,融合直线角度、长度信息,提取冬小麦分蘖个数。基于已开发的机器视觉设备,采集60幅自然光照条件下抽穗期冬小麦田间图像,经检测发现分蘖数平均识别精度达到93.3%,平均误检率为7.0%,平均漏检率为5.2%。结果表明:该算法可实现在自然光照条件下对抽穗期冬小麦田间图像的分蘖数准确识别,可为自动化、高通量田间信息采集系统设备的研发奠定一定的基础。
Tiller number is an important indicator for winter wheat yield evaluation and the high-throughput acquisition of tiller traits is of great significance for wheat breeding research and functional genome identification.In view of the current problems of wheat tiller number detection mainly relying on manual census,which is time-consuming,labor-intensive and poor repeatability,a rapid detection method of winter wheat tiller number at heading stage based on machine vision was proposed in this paper.First of all combining with morphological operations,we took super green algorithm and constant color segmentation in RGB color space to obtain the foreground target area and boundary.And then basing on wavelet transform haar feature offline feature detection extended template the plant edge information was extracted completely.Finally,the Hough line detection algorithm was used to fuse the angle and length information of the line to extract the number of winter wheat tillers.Experimental results showed that the identification algorithm proposed in this study had good robustness under field conditions.The average accuracy of winter wheat tiller number identification under natural comprehensive illumination reached 93.3%,the average false detection rate was 7.0%,and the average missed detection rate was 5.2%.The experiment proves that the research algorithm can realize the identification and detection of the number of tillers in the heading stage image under natural lighting conditions,which is helpful to lay a certain foundation for automated and high-throughput field information collection system equipment.
作者
张开
马淏
姬江涛
金鑫
朱旭
Zhang Kai;Ma Hao;Ji Jiangtao;Jin Xin;Zhu Xu(College of Agricultural Equipment Engineering, Henan University of Science and Technology, Luoyang 471003,China;Henan International Joint Laboratory of Intelligent Agricultural Equipment Technology, Luoyang 471003, China)
出处
《农机化研究》
北大核心
2022年第8期15-20,共6页
Journal of Agricultural Mechanization Research
基金
国家自然科学基金青年基金项目(61805073)
国家重点研发计划项目(2018YFD0700302-02)。
关键词
冬小麦
分蘖数
机器视觉
直线检测
颜色分割
超绿算子
HAAR变换
winter wheat
number of splits
machine vision
line detection
color segmentation
ultra-green operator
Haar transformation