摘要
为了解决常规农业移动机器人导航基准线提取方法存在识别速度慢、检测精度低以及对光照变化敏感等问题,提出1种自然光照环境下基于人工蜂群算法的视觉导航路径提取方法。首先,将视觉传感器获取的作物图像进行灰度化处理,通过图像熵对灰度图像质量进行估计,当光照条件变化时,在线调整摄像机曝光时间,使获取的图像质量达到最佳状态,以提高后续图像处理对光照变化的适应能力。然后,采用类间最大方差法对图像进行分割,将作物信息与土壤背景分离,运用形态学滤波方法消除分割图像中的杂草噪声。最后,对图像顶部和底部分别进行灰度垂直投影,获取作物行区域并提取作物行特征点,利用人工蜂群算法搜索2个特征点,使其构成的直线所含目标点数最多,并将这条直线作为作物行中心线,进而得到导航路径。结果表明,在不同光照度条件下,基于图像熵的曝光时间调整方法可以有效降低光照度变化对后续图像处理的影响;基于人工蜂群算法的导航基准线提取方法可以快速有效地识别作物行与导航路径,处理1幅640×480像素的图像平均耗时76.4 ms,与传统导航基准线提取方法(Hough变换算法、最小二乘法)相比,人工蜂群算法具有检测速度快、准确性高的特点。本研究提高了应用于田间作业的农业移动机器人导航路径识别精度。
To solve the problems such as slow recognition speed,low detection accuracy and sensitive to illumination variation in extraction methods for navigation reference lines of conventional agricultural mobile robots,a new navigation line extraction method based on artificial bee colony algorithm under natural light condition was proposed.Firstly,the crop images obtained by visual sensor were grayed,and image entropy was used to estimate the quality of the images.When the illumination condition varied,exposure time of the vidicon was adjusted online to make the quality of the obtained images in the best condition,so as to enhance the adaptability of further image processing to illumination variation.Secondly,the Otsu algorithm was used to separate the crop information from soil background,the redundant information of weeds in the binary images was eliminated by morphological filtering algorithm.Finally,grey vertical projection method was adopt on the top and bottom of the images to obtain the position of crop row and extract feature points.Two feature points were searched by artificial bee colony algorithm,and the formed straight line included the most target points and was used as the center line of crop rows,then the navigation path was obtained.The results showed that exposure time adjustment method based on image entropy could effectively reduce the effect of illumination variation on following image processing under different illumination conditions.The guidance line extraction method based on artificial bee colony algorithm could recognize crop rows and navigation path quickly and effectively.The time for processing an image with 640×480 pixels was about 76.4 ms.Compared with conventional guidance line extraction algorithms,the artificial bee colony algorithm showed the advantages of fast detection and high accuracy.The research improves the recognition accuracy of navigation path for agricultural mobile robots working in the fields.
作者
孟庆宽
杨晓霞
刘易
刘永江
张振仪
MENG Qing-kuan;YANG Xiao-xia;LIU Yi;LIU Yong-jiang;ZHANG Zhen-yi(College of Automation and Electrical Engineering, Tianjin University of Technology and Education, Tianjin 300222, China;Tianjin Key Laboratory of Information Sensing and Intelligent Control, Tianjin 300222, China)
出处
《江苏农业学报》
CSCD
北大核心
2020年第4期919-929,共11页
Jiangsu Journal of Agricultural Sciences
基金
天津市教委科研计划项目(JWK1613)。
关键词
农业移动机器人
机器视觉
导航线识别
图像熵
人工蜂群算法
agricultural mobile robot
machine vision
guidance line recognition
image entropy
artificial bee colony algorithm
