摘要
基于机器视觉的导航系统的目标是能自动采集农田环境图像,通过处理分析图像识别出导航路径,最终计算得出系统导航参数以控制农业机械沿导航路径行走。导航基准线的提取是机器视觉导航的基础。
本文主要研究了农业机械的视觉导航技术,实现了农田图像中导航基准线的提取。基于VFW技术开发了视频采集软件,使用USB接口的数字摄像头采集农田中作物行图像,能够实现实时的捕获图像。对色彩模型进行分析的基础上,选择使用RGB模型研究图像。利用2G-R-B法对彩色农田图像灰度化,使用中值滤波去除图像的噪声,对图像开运算去除孤立点和作物行中的孔洞,使用自适应阈值方法对图像二值化,并分析了区域生长法分割图像的适合场合,为后续处理提供条件。将导航定位点检测的方法分为边缘检测和中心线检测两类,使用Sobel算子对作物行边缘检测；在中心线检测时使用图像水平条分割,和垂直投影法确定作物行的导航定位点。提出了针对确定作物行提取定位点的方法,该方法根据垂直投影图的曲线波峰位置初步确定基准线位置,然后将感兴趣区域图像分成若干个水平条,对每个水平条用垂直投影法找出导航定位点。开发了导航线提取系统,并对三种方法的定位精度进行了对比。
本文主要针对视觉导航中精度、实时性不高的情况,改进了数据处理量大、耗时多和精度低的算法。首先为避免对作物行像素的漏检,改进了2G-R-B法判断像素值的条件,有效地分割作物行和背景,同时提高了灰度化的效率。改进了中值滤波的排序算法,使耗时减少一倍。提出了基于确定作物行提取定位点的方法,使用区域生长法分割确定作物行为感兴趣区域,大大的减少了无关信息量,提高了精度和实时性。
本文使用C++Builder6.0和OpenCV视觉库开发了导航基准线提取系统,对提取定位点的方法进行了分析。
本文基于确定作物行提取定位点的方法只针对中间作物行提取基准线,比对全部作物行提取基准线的算法不仅精度高,耗时减少2倍多,可以满足农业机械农田作业的需要。
The goal of the navigation system based on machinery vision is calculating the navigation parameters by analysed the farmland image which is acquired by the vision system to control agricultural machinery to walk along with the navigation path. The baseline extraction is the foundation of machine vision navigation.
This paper mainly studied the vision navigation technology of agricultural machinery which extracted the baseline of farmland image. Developed the software which can acquire image in real-time based on VFW technology and USB interface digital camera. Select the RGB model to study images based on the analysis of color models. Using2G-R-B method to grayscale image and median filter to remove image noise. Open operation was used to remove outliers and holes in the crop row image. Making use of adaptive threshold method to binary a gray image and analysis the suitable occasion of using the region growing segmentation method to providing conditions for the subsequent processing. Divided the method of detecting navigation logical points to two types, which is edge detection and centerline detection. The Sobel operator was used to detect the edge of the crop row of image. In the centerline detection, using horizontal splits segmentation and vertical projection to find the position of the navigation points. A new baseline detecting method based on the specific crop row was presented. It based on the curve peak of the vertical projection to initially determine the position of the baseline. Then divided the interested region to several horizontal splits, and using vertical projection to find the navigation points. Developed a baseline extraction system to contrast the accuracy of the three methods which was mentioned in the above,
In view of the situation of the low accuracy and speed, the algorithms which was time-consuming.and need large amounts of data processing was improved. The2G-R-B method was improved to avoid missing the important pixels of crop row image. Then it can split the crop rows and background effectively and improved the efficiency of graying. The sorting algorithm of median filter was improved to reduce the time consumption. A new baseline detecting method based on the specific crop row was presented. It used the region growing method to extract the specific crop row as the interest region to reduce the useless information and improve the accuracy and speed in processing image.
In the article a navigation baseline extraction system was developed making use of C++Builder6.0and OpenCV vision library, and it analysed the method of extraction positioning points.
In the article the baseline extraction method only for the middle crop row based on the method of navigation points extraction in specific crop row has high accuracy and reduces twice as time-consuming. So the new method can meet the requirements of agricultural operations of agricultural machinery.
引文
[1]杨为民,李天石,贾鸿社.农业机械机器视觉导航研究[J].农业工程学报.2004,20(1)：160-165.
[2]高峰,李艳等.基于遗传算法的农业移动机器人视觉导航方法[J].2008,39(6)：127-131.
[3]Marchant JA. Tracking of row structure in three crops using image analysis[J].Computers and Electronics in Agriculture,1996(15):161-179.
[4]J.F.Reid et al. Agricultural automatic guidance research in North America[J].Computers and Electronics in Agriculture.2000(25):155-167.
[5]Marchant JA. Tracking of row structure in three crops using image analysis[J]. Computer Electron Agric 1996:161-79.
[6]Marchant JA, Brivot R. Real-time tracking of plant rows using a hough transform. Real-Time Image 1995:363-71.
[7]Toru Torii. Research in autonomous agriculture vehicles in Japan[J]. Computer and Electronics in Agriculture,2000,25(1):133-153.
[8]Wilson J N.Guidance of agricultural vehicles-a historical erspective[J].Computers and Electronics in Agriculture,2000,25:3-9.
[9]S.Han, Q. Zhang,B.Ni,J.F.Reid A guidance directrix approach to vision-based vehicle guidance systems[J]. Computers and Electronics in Agriculture 43(2004)179-195.
[10]M.Kise, Q.Zhang, F.Rovira Mas.A Stereo vision-based Crop Row Detection Method for Tractor-automated Guidance[J]. Biosystems Engineering (2005) 90(4),357-367.
[11]Eric Royer, Maxime Lhuillier, Michel Dhome and Jean-Marc Lavest. Monocular Vision for Mobile Robot Localization and Autonomous Navigation[J]. International Journal of Computer Vision 74(3),237-260,2007
[12]Raphael Linker, Tamir Blass. Path-planning algorithm for vehicles operating in-orchards[J]. Biosystems Engineering 101(2008),152-160.
[13]熊利荣,丁幼春等.机器视觉技术在农业领域中的应用[J].湖北农机化.2004(4)：23.
[14]邓继忠,张泰岭,石江.机器视觉技术在农业机械中的应用[J].农机化研究.2001：91-93.
[15]沈明霞.自主行走农业机器人视觉导航信息处理技术研究[D].南京：南京农业大学,2001.
[16]王荣本,纪寿文,初秀民等.基于机器视觉的玉米是非智能机器系统设计概述[J].农业工程学报.2001,17(2)：151-154.
[17]任永新,谭豫之,杨会华等.基于模糊控制的黄瓜采摘机器人视觉导航[J].江苏大学学报.2009,30(4)：343-346.
[18]袁佐云,毛志怀,魏青.基于计算机视觉的作物行定位技术[J].中国农业大学学报,2005,10(3)：69-72.
[19]虢莉敏,陈宁,刘宏刚等.基于DirectShow技术的视频采集方案的实现[J].电子技术,2007(2)：66-69.
[20]谢志鹏,陈锻生.基于VFW的实时视频图像采集处理程序设计[J].微机发展.2004,14(11)：121-123.
[21]王铬.回调函数在软件设计中的应用[J].河南教育学院学报.2003(3)：44-46.
[22]王集成.基于VFW图像采集应用[J].计算机与数字工程.2009,37(7)：136-139.
[23]任重远.显示器相关颜色空间RGB与无关颜色空间Lab的关系[J].今日印刷.2007(6)：44-45.
[24]吴富宁,朱虹,郑丽敏等.计算机辅助小麦图像识别应用中颜色辅征基本参量的表达[J].农业网络信息,2004,4：10-14.
[25]Woebbecke D M,Meyer G E,Von bargen K.Color indices for weed identification under various soil,residual,and lighting conditions [J].Transactions of the ASAE,1995,38(1);259-269
[26]王俊芳,王正欢,王敏.常用图像去噪滤波方法比较分析[J].2009(16)：310-311.
[27]刘丽梅,孙玉荣,李莉.中值滤波技术发展研究[J].云南师范大学学报.2004(4)：23-27.
[28]李敏,蒋建春.基于腐蚀算法的图像边缘检测的研究与实现[J].计算机应用与软 件.2009(1)：82-84.
[29]张莹.开闭运算在消除图像噪声中的应用研究[J].潍坊学院学报.2002(2)：65-66.
[30]阴国富.基于阈值法的图像分割技术[J].现代电子技术.2007(23)：107-108.
[31]Otsu N.A threshold selection method from gray level histograms[J].IEEE Trans.On Systems Man and Cyberm.1997,9(1):62-66.
[32]张建光,李永霞.基于区域的图像分割[J].科技资讯.2011(26)：13.
[33]吴佳艺.基于机器视觉的农林环境导航路径生成算法研究[D].杭州：浙江工业大学.2009.
[34]康乾.常用边缘检测方法分析与比较[J].信息与电脑.2010(01)：139.
[35]张卫,杜尚丰.机器视觉对农田中定位基准线的识别[J].中国农业大学学报,2006,11(4)：75-77.
[36]卢卫娜.车辆视觉导航方法研究[D].西安：西北工业大学.2006.
[37]H.T.Sogaard, H..J. Olsen. Determination of crop rows by image analysis without segmentation[J].Computers and Electronics in Agriculture 38(2003)141-158.
[38]任永新,谭豫之,杨会华等.基于模糊控制的黄瓜采摘机器人视觉导航[J].江苏大学学报.2000.30(4)：343-346.
[39]刘建友,李宝树等,航拍绝缘子图像的提取和识别[J].传感器世界.2009(12)：22-24.
[40]商飞,王丰贵,田地等.一种基于圆内接直角三角形的圆检测方法[J].光学学报,2008,28(4)：739-743.
[41]赵颖,王书茂,陈兵旗.基于改进Hough变换的公路车道快速检测算法[J].中国农业大学学报.2006,11(3)：104-108.
[42]秦小文,温志芳,乔维维.基于OpenCV的图像处理[J].电子测试.2011(7)：39-41.