摘要
杂草对农作物的侵害一直是人们关注的问题，目前除草的主要方法是利用除草剂，它不仅去除农作物中杂草，同时还提高了作物产量。但是由于除草剂过度使用，一方面增加了生产投入成本，另一方面也造成了环境的污染。为适应精细农业的要求和需要，本文对杂草识别进行初步研究。
本研究利用图像处理技术对复杂土壤背景环境下杂草进行识别处理，本系统主要包括三部分。首先是利用图像的颜色特征增加绿色植物与土壤的对比度，把植物从复杂背景环境下分离出来，转换真彩色图像为灰度图像。其次是对图像的阈值分割算法进行研究，选择合适的分割算法，即可变阈值分割法，它能比较稳定、不失真的把灰度图像转换为二值图像。第三部分则是在前两步的基础上建立了一个杂草自动识别系统，该系统能识别出图像中的杂草，并根据杂草的位置定点模拟喷洒。本系统主要利用图像处理中的形态学算法来实现，对植物和杂草在形状上有明显区别情况能得到较好的识别效果。整个系统是基于Windows2000操作系统平台下应用VC++6.0开发的，VC++语言在算法的处理速度上较其它语言快，适合实时处理的需要。
本文在研究过程中对多种算法进行比较，并采用邻域滤波法和快速中值滤波法对原始图像和灰度图像滤波，去除各种噪声对图像的影响。同时还对与图像方向和大小无关的一些形状特征参量进行研究，为进一步对杂草的识别研究做准备，进而完善杂草识别系统。
图像的颜色特征与形状特征的结合使用，在杂草的识别处理研究中起着重要的作用，并且这种方法的可行性在本研究中得到了验证。
杂草识别系统的研究对合理使用除草剂，增加农业的自动化水平，保护环境具有一定的现实意义。
Weed infestation to crop is a question which people pay attention to all the while, at present main weeding method is applying herbicide, which not only can weed in fields, but also advance yield of crop. But for over-applying herbicide, on the one hand which increases cost of production, on the other hand which causes pollution to environment. To adapt to the request and need of precision agriculture, the paper fundamentally researched for weed identification.
The research applied image processing technology to identify weed under complex soil background, the system included three parties. First, applying color feature increased the . contrast between green plants and soil, separated the plants from complex background and changed real color image to gray-level image. Secondly, Some algorithms of image threshold segmentation were researched and selected a suitable segmentation method, namely changing threshold segmentation, the algorithms could transfer gray-level image into binary image with stabilization and no distortion. Thirdly, a automatic identification system of weed wae built based on former two steps, which might identify weeds in image and carries out simulative spay to site-specific weed locations. The system was realized by morphological algorithm of image processing, which could obtain a better identification effect for some plant and weeds that are obviously different in figure. The whole system was developed through applying VC++6.0 language under t
he Windows2000 operating system, the processing speed of VC language is quick than the other language, so it fits for real-time processing request.
The paper compared many algorithms during research, applied neighborhood filtering method and fast median filtering method to filter original images and gray-level images and eliminated effects of all kinds of noise to images. Some shape features parameters were also researched, these parameters are not correlated to image direction and image size, these researches would prepare for father identification research of weed and thereby perfect weed identification system.
The approach that image color feature and shape feature was combined was important in field of weed identification, feasibility of the method has been validated in the research.
The research of weed identification system had practical significance for suitable applying herbicide, enhancing automatic level of agriculture and protecting environment.
Candidate: Liu Zhenheng Major: Agricultural electrization and automatization Supervisor: Prof .Zhang Changli
引文
1 陈纯．计算机图像处理技术与算法．清华大学出版社，2003：119-125
2 崔屹．图像处理与分析——数学形态学方法及应用。科技出版社，2003：30-42
3 方如明，蔡健荣等．计算机图像处理技术及其在农业工程中的应用．清华大学出版1999：59-64
4 何斌，马天予等．Visual c++数字图像处理．人民邮电出版社，2002
5 何东健，杨青．实用图像处理技术．陕西科技出版社，1998：54-56
6 候俊杰．深入浅出MFC．华中科技大学出版社，1998
7 李孙荣．杂草及其防治．北京农业大学出版社，1991：90-95
8 李光熙．除草剂的开发，研制与应用．杂草科学，1998：(1)2-3
9 刘志敏，杨杰等．数学形态学的图像分割算法．计算机工程与科学，1998，(4)：21-27
10 马承中，刘滨等．农田杂草识别及防除．中国农业出版社，1999：247-281
11 马奇祥，常中先等．农田杂草化学防除图谱．河南科学技术出版社，2001：1-24
12 阮秋琦．数字图像处理学．电子工业出版社，2001：429-453
13 容观澳．计算机图像处理．清华大学出版社，2000：129-131
14 孙兆林．MATLAB 6．x图像处理．清华大学出版社，2002：267-282
15 王华，叶爱亮．VC++6．0编程实例与技巧．机械工业出版社，1999
16 王新成．高级图像处理技术．中国科学技术出版社，2001：18-20
17 王郑耀．数字图像的边缘检测．西安交通大学，2003：16-22
18 章毓晋．图像处理和分析．清华大学出版社，1999：82-93
19 朱志刚．数字图像处理．电子工业出版社，1999：50-60
20 David J.Kruglinski.Visual C++技术内幕．潘爱民，王国印．清华大学出版社，1999
21 Microsoft公司著．Visual C++6.0类库参考手册．北京希望电脑公司，1999
22 Milan Sonka, Vaclav Hlavac.Image processing, analysis, and machine vision.人民邮电出版社，2002:257-261
23 Rafael C．Gonzalez，Ri chard E．Woods．数字图像处理．阮秋琦，阮宇智等译．第二版．电子工业出版社，2003：224-240
24 纪寿文，王荣本等．应用计算机图像处理技术识别玉米苗期田间杂草的研究．农业工程学报．2001，(2)：154-156
25 李小文．利用拉普拉斯—高斯模板进行边缘检测．华南师范大学学报(自然科学版)．1997(2)：53-55
26 吕朝辉，陈晓光等．机器视觉田间植物检测与识别技术．吉林工业大学自然科学学报．2001，(3)：90-94
27 王坤明，朱双东等．自动选取阈值方法比较研究．抚顺石油学院学报．2002，Vol22(2)：70-72
28 王荣本，纪寿文等．基于机器视觉的玉米施肥智能机器系统设计概述．农业工程学报．2001，(3)：151-153
29 苑玮琦等．一种基于梯度极值的边缘检测算法．信息与控制．1997，Vol．2(2)：117-120
30 周文等．计算机图像处理技术在烤烟烟叶形状特征提取中的应用．计算机应用．2000(1)：12-14
31 Aubtale A S. Automatic threshold of gray-level pictures using two-dimensional entropy. Computer Vision ,Graphics and image processing. 1989, Vo147:23-33
32 C.-C. Yang, S.O. Prasher. Recognition of weeds with image processing and their use with fuzzy logic for precision farming. Canadian agriculture engineer. 2000, Vo142(4):195-200
33 Dr.R. Tsheko. Discrimination of plant species using co occurrence matrix of leaves. The DIGR journal of scientific research and development. 2002, Vol (4): 1-12
34 D.M. Woebbecke, G. E.Meyer. Color indices for weed identification under various soil, residue, and lighting conditions. Transactions of the ASAE. 1995, Vol38 (1), 259-269
35 D.M. Woebbecke, G.E. Meyer. Shape features for identifying young weeds using image analysis. Transactions of the ASAE. 1995, Vol38 (1), 271-281
36 Franz. Shape description of completely visible and Partially occluded leaves for identifying piants in digital images. ASAE, Paper No. 90:40-70
37 Franz. E, M.R. Gebhardt et al..The use of local spectral properties of leaves as aid for identify weed seeding in digital images. Transactions of the ASAE. 1991, Vol34 (20): 682-687
38 G.E.Meyer T.Mehta. Textural imaging and discriminate analysis for distinguishing weeds forspotspraying. Transaction of the ASAE. 1998,VoI41(4):1189-1197
39 Joseph. Identifying effective criteria for weed detection using machine vision. ASAE Paper. 1992, No. 97-3134
40 Kincaid D.T, R.B. Schneider. Quantification of leaf shape with a microcomputer and fourier transform.CanadianJ. of Botany.1983, 61:2333-2342
41 L. Tian, J.F. Reid etc. Development of a precision sprayer for site-specific weed management. Transaction of the ASAE. 1999, Vol 42(4):839-900
42 L. tang, L. Tian. Color image segmentation with genetic algorithm for in-field weed sensing. Transaction of the ASAE. 2000, Vol 43(4):1019-1027
43 M.S. El-Faki, N. Zhang. Factors affecting color-based weed detection.Transaction of the ASAE. 2000, Vol 43(4):1001-1009
44 N. Zhang, C. Chaisattapagon. effective criteria for weed identification in wheat fields using machine vision, Transaction of the ASAE. 1995, Vol 38(3):965-974
45 R.D. lamm, D.C. Slaughter. Precision weed control system for cotton.Transaction of the ASAE. 2OO2, Vol 45(1):231-238
46 S.A. Shearer, R.G. Holmes. Plant identification using color co-occurrence matrices Transaction of the ASAE. 1990, Vol 33(6):2037-2044
47 T.F. Burks, S.A. Shearer. Classification of weed species using color texture features and discriminate analysis. Transaction of the ASAE. 2000, Vol 43(2):441-448
48 T.F. Burks, S.A. Shearer. Backpropagation neural network design and evaluation for classifying weed species using color image texture. Transaction of the ASAE. 2000, Vol 43(4): 1029-1037
49 Zhang. N,C. Chaisattapagon. Effective criteria for weed identification in wheat fields using machine vision. Transactions of the ASAE. 1995,38(3):956-974
50 http://www.agrionline.net.cn/new_agri/list.asp?id=2644