基于嵌入式机器视觉的玉米苗分级检测系统设计  

Design of Corn Seedling Classification Detection System Based on Embedded Machine Vision

作　　者：赵献立 王志明[1] 田鹏飞 Zhao Xianli;Wang Zhiming;Tian Pengfei(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China;Ecovacs Robotics Technology Co.,Ltd.,Suzhou 215100,China)

机构地区：[1]南京理工大学机械工程学院,南京210094 [2]科沃斯机器人科技有限公司,江苏苏州215100

出　　处：《农机化研究》2023年第9期66-70,共5页Journal of Agricultural Mechanization Research

基　　金：国家重点研发计划项目(2017YFD0701500,2017YFB0304700,2018YFD0700800)。

摘　　要：与传统的PC式机器视觉系统相比,嵌入式机器视觉系统具有低功耗、空间占比小、操作简单、易拓展及成本低的特点,在图像采集和处理方面应用越来越广泛。为此,以ARM Cortex A9为硬件平台,结合嵌入式Linux系统,通过USB摄像头采集图像并运行图像处理算法,设计了交互软件系统,实现了玉米苗的分级检测。研究结果显示:使用本算法能够有效地对玉米幼苗进行提取,分割精度达到93.87%,处理后分级结果显示正确,运行时间稳定在240ms左右,系统性能稳定,具有一定的应用价值。Compared with traditional PC-type machine vision systems,embedded machine vision systems have the characteristics of low power consumption,small space,simple operation,easy expansion,and low cost.Therefore,embedded machine vision systems are widely used in the image acquisition and processing aspects.In this paper,ARM Cortex A9 is used as the hardware platform,combined with the embedded Linux system,the image is collected through the USB camera and the image processing algorithm is deployed,and the interactive software system is designed to realize the grading detection of corn seedlings.The results show that using this algorithm can effectively extract corn seedlings,and the segmentation accuracy reaches 93.87%.The classification results are displayed correctly,the running time is at about 240ms stably,the system performance is stable,and it has broad application value.

关 键 词：嵌入式LINUX系统 图像处理 分级检测 玉米苗 

分 类 号：S24[农业科学—农业电气化与自动化] S237[农业科学—农业工程]