摘要
杂草会给农业生产造成严重的损失,化学除草操作简单,效率较高,但面临在粮食和环境中残留所引起的安全问题。变量喷药能够根据杂草在田间的位置、种类和密度调节除草剂喷洒的时机和剂量,但要以对田间杂草的准确识别为前提。为此,开发了一种利用计算机视觉的玉米田间变量喷药除草系统,利用相机拍摄田间图像,导入计算机中进行分析,提取杂草信息后生成处方图传递给喷药控制系统;喷药控制系统根据处方图,结合机械的速度控制喷头的开闭时机和程度,实现变量喷药。系统对田间杂草的识别率为90%,处理单张图片的平均耗时为86ms,能满足对杂草进行实时识别的要求。在室内试验中,机械行进速度为0.3m/s时,系统具有很好的准确性和作业效率。
Weeds can cause serious damage to agricultural production,chemical herbicides are simple and efficient,but also face the food and the environment caused by the residual safety problems. Variant spraying can adjust the timing and dose of herbicide spraying according to the position,species and density of weeds in the field,but on the premise of accurate identification of field weeds. In this paper,we developed a maize field variable spraying herbicidal system using computer vision. We used the camera to photograph the field image,import it into the computer for analysis,and extract the weed information to generate the prescription map and transmit it to the spray control system. Spraying control system according to the prescription,combined with the speed of the machine to control the opening and closing of the nozzle and the extent to achieve variable spray. The rate of recognition of field weeds is 90%,and the average time taken to process a single picture is 86 ms,which satisfies the requirement of real-time identification of weeds. In the indoor test,the mechanical travel speed of 0. 3m/s,the system has the best accuracy and operating efficiency.
出处
《农机化研究》
北大核心
2018年第3期217-220,225,共5页
Journal of Agricultural Mechanization Research
基金
广西高校中青年教师基础能力提升项目(KY2016YB685)
