摘要
为了探索种薯自动化切种过程,填补关于马铃薯芽眼识别的研究空白,提出一种基于机器视觉技术的芽眼识别方法:从摄像头采集到马铃薯图像后进行计算机图像处理,从彩色空间中利用欧式距离直接分割芽眼区域,在灰度空间中对图像进行中值滤波后利用模糊技术对图像进行增强,之后利用动态阈值分割法分割芽眼区域,结合两个空间的分割结果后利用数学形态学处理方法标记出芽眼。结果显示:在彩色空间中,芽眼识别准确率为62%;在灰度空间中,识别率达到89%。将二者有机结合后,获得了96%的识别准确率。该方法识别成功率高,鲁棒性强,且芽眼区域标记完整,可为种薯切种自动化奠定基础。
In order to explore automatic seed cutting of potato and potato bud eye recognition,the present paper reported a relevant approach based on machine vision. It used camera to collect images,and computer to process images. In the part of image segmentation,it utilized Euclidean distance to segment the region of bud eye in color space.Then,the dynamic threshold segmentation was carried out after image filtering and intensification,to segment bud eyes in gray space. At last,it combined the two segmentation part,and used mathematical morphology to label bud eye out. It was shown that in color space,the recognition accuracy rate was 62%,and in gray space,the recognition accuracy rate was 89%. Combined together,the recognition accuracy rate was 96%. The proposed method exhibited high recognition accuracy rate,acceptable robustness and intact boundary of bud eye,which laid theoretical foundation for automatic seed cutting of potato.
出处
《浙江农业学报》
CSCD
北大核心
2016年第11期1947-1953,共7页
Acta Agriculturae Zhejiangensis
基金
国家自然科学基金(61462059)
甘肃省自然科学基金(148RJZA054)
关键词
芽眼识别
机器视觉
图像分割
马铃薯
bud eye recognition
machine vision
image segmentation
potatoes
