基于图像特征的玉米叶部病害模糊识别研究与应用
|
摘要
随着计算机处理能力的不断增强,以及图像处理与识别技术的快速发展,数字图像处理与识别技术在农业中的应用越来越广泛,并将成为实现农业信息化与自动化的重要技术力量农业信息采集工作量巨大,信息的实时性和准确性是农业生产和科学研究领域普遍关注的问题,如何及时快速地进行农作物病害的准确判断一直是计算机技术面向农业领域研究工作的一项重要内容作物的病虫草害严重影响作物的产量和质量,本文针对作物病虫草害的自动化识别程度低,诊断不准确的问题,以计算机图像处理技术为重要手段,综合运用图像处理和植物病理学方面的知识,在国家自然科学基金项目(303060047)资助下,以常见的玉米叶部病害为研究对象,提出可行的识别方法,提高识别诊断的精度,为作物病虫草害自动识别诊断的相关研究提供理论依据
本论文阐述了图像处理与识别的基本理论,使用VC++编程语言建立了玉米叶部病害的诊断识别软件系统课题组选择图像获取设备,搭建硬件检测系统,获取大田环境下的玉米叶部病害图片研究根据玉米叶部病害的特点,综合应用图像平滑阈值法区域标记局部阈值法和区域增长法相结合的算法(TSRG),对玉米叶部病害图片进行分割统计病斑的个数出去冗余斑点,同时提取出病斑的颜色和形状特征；采用基于模糊决策最大隶属度原则的模糊识别算法对玉米叶部的大斑病小斑病灰斑病褐斑病弯苞菌叶斑病锈病六种病害进行分类识别,并综合与其它识别方法的识别结果进行比较,得到较高精度的识别诊断结果,为研究作物病虫草害的智能识别诊断提供了软件和技术支持
本研究取得了以下两方面的进展：第一,实现了玉米叶部大斑病小斑病灰斑病褐斑病弯苞菌叶斑病锈病六种病害的自动识别,识别准确率达到90%以上；第二,分析了利于玉米叶部病害识别的病斑特征,将模糊识别算法应用于玉米叶部病害识别适应玉米病害的特点,获得较高的识别精度同时综合其它的识别算法得出模糊识别应用的优势,提高了识别的准确性和可靠性
将机器视觉应用于玉米叶部病害的识别诊断,拓展了机器视觉的应用范围,为机器视觉技术在农业领域的研究应用奠定了基础
Along with increasing ability of computer process as well as rapid development of image processing and recognition technology, this technology not only is applied to agriculture more and more widely, but also is an important technology to achieve agriculture information and automation. The amount of work for gathering agriculture information is tremendous; besides the realtime and accuracy of information are widely paid attention to in agriculture production and scientific research. How to diagnose disease of crops timely and accurately is always an important part of research facing to the agriculture field through computer technology. At the same time crop disease, insect pest and weed affect the output and quality of crop badly. The paper mainly aimed at the problem of low precision and untimely diagnosis about crop disease, by means of computer image processing technology as an important means, took common maize leaf diseases as study object and brought forward feasible method to improve diagnosis precision applying the knowledge on image processing and plant physiology synthetically, therefore it provides theory to the correlative research of auto recognition and diagnosis of crop disease, insect pest and weeds.
The paper represented the theory of image processing and recognition, made use of VC++ programming language build up the recognition software system and project team built the hardware system to get the diseases image. According to the characteristics of maize leaf disease, TSRG method was adopted to segment the disease image; area-marking and Freeman link code method were used calculating the num of disease well as wiping off redundancy dots. And then extracted the disease spot's color and form feature and saved in the database. Based on fuzzy decision making and mean of maximum method, Fuzzy Recognition algorithm was used to recognize six kinds of maize leaf disease. The paper integrated the results of other recognition methods, which were used to research crops, and got the effective identification of diagnostic results, which provides software and technology support for crops'diseases, pests and weeds diagnosis and recognition research.
The study made two aspects of progress in the following. First, it realized automatic identification of the common six kinds of maize leaf disease, and the recognition accuracy rate of over 90%. Secondly, the paper analysised the feature of the maize leaf disease and selected the features which made a greater contribution to recognition; at the same time, Fuzzy Recognition algorithm adopted to maize leaf disease recognition was a great breakthrough for crops Recognition Research. Because of it integrated the advantages of the other methods and got the identification of diagnostic results, it improved the accuracy and reliability of recognition.
The research applied computer vision technology in maize disease recognition, which extends the application area of machine vision and also provide a use for reference in the field of agriculture technology for machine vision.
本论文阐述了图像处理与识别的基本理论,使用VC++编程语言建立了玉米叶部病害的诊断识别软件系统课题组选择图像获取设备,搭建硬件检测系统,获取大田环境下的玉米叶部病害图片研究根据玉米叶部病害的特点,综合应用图像平滑阈值法区域标记局部阈值法和区域增长法相结合的算法(TSRG),对玉米叶部病害图片进行分割统计病斑的个数出去冗余斑点,同时提取出病斑的颜色和形状特征；采用基于模糊决策最大隶属度原则的模糊识别算法对玉米叶部的大斑病小斑病灰斑病褐斑病弯苞菌叶斑病锈病六种病害进行分类识别,并综合与其它识别方法的识别结果进行比较,得到较高精度的识别诊断结果,为研究作物病虫草害的智能识别诊断提供了软件和技术支持
本研究取得了以下两方面的进展：第一,实现了玉米叶部大斑病小斑病灰斑病褐斑病弯苞菌叶斑病锈病六种病害的自动识别,识别准确率达到90%以上；第二,分析了利于玉米叶部病害识别的病斑特征,将模糊识别算法应用于玉米叶部病害识别适应玉米病害的特点,获得较高的识别精度同时综合其它的识别算法得出模糊识别应用的优势,提高了识别的准确性和可靠性
将机器视觉应用于玉米叶部病害的识别诊断,拓展了机器视觉的应用范围,为机器视觉技术在农业领域的研究应用奠定了基础
Along with increasing ability of computer process as well as rapid development of image processing and recognition technology, this technology not only is applied to agriculture more and more widely, but also is an important technology to achieve agriculture information and automation. The amount of work for gathering agriculture information is tremendous; besides the realtime and accuracy of information are widely paid attention to in agriculture production and scientific research. How to diagnose disease of crops timely and accurately is always an important part of research facing to the agriculture field through computer technology. At the same time crop disease, insect pest and weed affect the output and quality of crop badly. The paper mainly aimed at the problem of low precision and untimely diagnosis about crop disease, by means of computer image processing technology as an important means, took common maize leaf diseases as study object and brought forward feasible method to improve diagnosis precision applying the knowledge on image processing and plant physiology synthetically, therefore it provides theory to the correlative research of auto recognition and diagnosis of crop disease, insect pest and weeds.
The paper represented the theory of image processing and recognition, made use of VC++ programming language build up the recognition software system and project team built the hardware system to get the diseases image. According to the characteristics of maize leaf disease, TSRG method was adopted to segment the disease image; area-marking and Freeman link code method were used calculating the num of disease well as wiping off redundancy dots. And then extracted the disease spot's color and form feature and saved in the database. Based on fuzzy decision making and mean of maximum method, Fuzzy Recognition algorithm was used to recognize six kinds of maize leaf disease. The paper integrated the results of other recognition methods, which were used to research crops, and got the effective identification of diagnostic results, which provides software and technology support for crops'diseases, pests and weeds diagnosis and recognition research.
The study made two aspects of progress in the following. First, it realized automatic identification of the common six kinds of maize leaf disease, and the recognition accuracy rate of over 90%. Secondly, the paper analysised the feature of the maize leaf disease and selected the features which made a greater contribution to recognition; at the same time, Fuzzy Recognition algorithm adopted to maize leaf disease recognition was a great breakthrough for crops Recognition Research. Because of it integrated the advantages of the other methods and got the identification of diagnostic results, it improved the accuracy and reliability of recognition.
The research applied computer vision technology in maize disease recognition, which extends the application area of machine vision and also provide a use for reference in the field of agriculture technology for machine vision.
引文
[1]方如明,蔡健荣,徐俐.计算机图像处理技术及其在农业工程中的应用.第一版.北京：清华大学出版社,1999.1-181.
[2]Casady W W, Paulsen M R. An Automated Kernel Positioning Device for Computer Vision Analysis of Grain.Transactions of the ASAE.1989.32(5)
[3]Sarkar N, R.R.Wolfe. Feature Extraction Techniques for Sorting tomatoes by Computer Vision.Translations of the ASAE.1985.28(3):970-974.
[4]Miller.B.Kand, Deiwiche.M.J. A Color Vision System for Peach Grading. Translation of ASAE.1989.32(4):1484-1490.
[5]Sherer.S A, Payne F A. Color and Defect Sorting of Bell Pepper Using Machine Vision.Translations of the ASAE.1990.38(3):959-964.
[6]Tao Y, Heinemann P H, Varghese Z, et al. Machine vision for color inspection of potatoes and apples.Trans of the ASAE.1995.38(5):1555-1561.
[7]Humpheries S. Identification of Plant Parts Using Color and Geometric Image Data.Translations of the ASAE.1993.36(5):1493-1500.
[8]Shimizu H, Heins R D. Computer vision based system for plant growth analysis. Trans of the ASAE.1995.38(3):959-964.
[9]林开颜,徐立鸿,吴军辉.计算机视觉技术在作物生长监测中的研究进展.农业工程学报.2004.20(4)：279-282.
[10]陈顺三.奶牛体形线性评定的图像处理技术研究.北京农业工程大学博士论文,1996.25-40
[11]张建平.数字图像处理在农产品质量检测中的应用一烟叶质量的自动分级.江苏理工大学博士论文,1994.55-65.
[12]刘禾.计算机视觉在水果自动分级中的应用.中国农业大学博士论文,1994.58-70
[13]宋韬.基于计算机图像处理的作物籽粒的识别研究.中国农业大学博士论文,1994.69-80
[14]杨秀坤,陈晓光,马成林等.用遗传神经网络方法进行苹果颜色自动检测的研究农业机械学报.1997.13(2)：76-173.
[15]陈佳娟,纪寿文等.基于遗传神经网络的玉米叶色的自动测定研究.农业工程学报.2000.15(5)：115-117.
[16]毛罕平,徐贵力,李萍萍.番茄缺素叶片的图像特征提取和优化选择研究.农业工程学报.2003.19(2)：133-136.
[17]张香琴,方如明,汪萍等.家蚕微粒子病的图像识别技术研究.农业机械学报.2001.32(5)：65-68
[18]李长缨,腾光辉,赵春江等.利用计算机视觉技术实现对温室植物生长的无损监测.农业工程学报.2003.19(3)：140-143.
[19]田有文,李成华.基于统计模式识别的植物病害彩色图像分割方法.吉林大学学报,2004.4,34(2)：291-293.
[20]马骏,王建华.一种基于数学形态学的植物病虫识别方法.深圳大学学报,2004,21(1)：72-75.
[21]Young J Han, Virgil L Quisenberry. Image Analysis to Quality Soil Pore Geometry. ASAE Paper, No.89-3004.
[22]Liao K, Paulsen M R, Reid J F. Real-time detection of color and surface defects of maize kernels using machine vision. Journal of Agriculture Engineering Research.1994,59:263～271.
[23]Shimizu H, Heins R D. Computer vision based system for plant growth analysis. Trans of the ASAE.1995,38(3):959～964.
[24]纪寿文,王荣本,陈佳娟等.应用计算机图像处理技术识别玉米田间杂草的研究[J].农业工程学报,2001,17(2)：154-156.
[25]吕俊伟,马成林,左春栓.用彩色图像识别田间洋葱和杂草的研究[J].农业工程学报,2001,17(5)：153-158.
[26]徐黎.VC++下的数字图像处理时限与研究[J].电脑开发与应用,2001,(4)：31-33.
[27]白耀辉,陈明,王举群.利用朴素贝叶斯方法实现异常检测.计算机工程与应用.352005,131-132.
[28]肖波,徐蔚然.基于贝叶斯分类器的混排文字切分与分类.计算机工程与应用.102005,83-85.
[29]吴立增,朱永利,苑津莎.基于贝叶斯网络分类器的变压器综合故障诊断方法.电工技术学报.20(4)2005,45-51.
[30]张少中,王秀坤,孙莹光.贝叶斯网络及其在决策支持系统中的应用.计算机工程.30(10)2004,1-3.
[31]陈晓光,于海业等.应用图像处理技术进行蔬菜苗特征量识别.农业工程学报.11(4)1995,23-26.
[32]安冈善文.图像处理技术在环境中的应用.电气学会杂志特集.1985,455-458.
[33]穗波雄信.根据图像提取植物的生长信息.农业机械学会关西支部第6次支部研究资料.10,1989,1-2.
[34]Bai Zhongying, Li Shaokun. The application of recognizing population feature of wheat images based on BP neural network. Proceedings of International Symposium on Future Software Technology, ISFST'2002. October 23-25,2002 231-236.
[35]Antonio J, et al. An evaluation of the application of infrared thermalimaging to the tele-diagnosis of sarcoptic mangein the Spanish ibex (Capra pyrenaica) Veterinary Parasitology. 109,2002,111-117.
[36]Cheng H D, et al. A novel fuzzy entropy approach to image enhancement and thresholding Signal Processing.75,1999,277-301.
[37]Cheng H D, et al. A Hierarchical Approach to Color Image Segmentation Using Homogeneity. IEEE Trans, on Image Processing.9(12)2000,2071-2082.
[38]Granitto, et al. Large scale investigation of weed seed identification by machine vision Computer and Electronics in Agriculture.2004.
[39]Ong S H, et al. Segmentation of color images using a two-stage self-organizing network Image and Vision Computing.20,2002,279-289.
[40]钱玲飞,刘玉树,李侃.朴素贝叶斯分类器在地形评估中的应用办法.计算机工程与应用.12.2005,189-191.
[41]王拓,蒋志雄,王颖初等.基于机器视觉的地膜下方棉苗的识别与定位.计算机工程与应用.24,2005：215-217
[42]钱玲飞,刘玉树,李侃.朴素贝叶斯分类器在地形评估中的应用办法.计算机工程与应用.12.2005,189-191.
[43]章毓晋.图象处理和分析.清华大学出版社.1999：219-220.
[44]李文化.模糊聚类新算法与模糊聚类神经网络.西安电子科技大学博士论文,1995.35-79.
[45]裴继红.基于模糊信息处理的图像分割方法研究.西安电子科技大学博士文,1998.50-89.
[46]吕铁英,彭嘉雄.图像轮廓特征提取新方法研究.中国图象图形学报.1999.4(8)：655-658.
[47]孙即样,王晓华,钟山等.模式识别中的特征提取与计算机视觉不变量.第一版.北京：国防工业出版社,2001.21-43.
[2]Casady W W, Paulsen M R. An Automated Kernel Positioning Device for Computer Vision Analysis of Grain.Transactions of the ASAE.1989.32(5)
[3]Sarkar N, R.R.Wolfe. Feature Extraction Techniques for Sorting tomatoes by Computer Vision.Translations of the ASAE.1985.28(3):970-974.
[4]Miller.B.Kand, Deiwiche.M.J. A Color Vision System for Peach Grading. Translation of ASAE.1989.32(4):1484-1490.
[5]Sherer.S A, Payne F A. Color and Defect Sorting of Bell Pepper Using Machine Vision.Translations of the ASAE.1990.38(3):959-964.
[6]Tao Y, Heinemann P H, Varghese Z, et al. Machine vision for color inspection of potatoes and apples.Trans of the ASAE.1995.38(5):1555-1561.
[7]Humpheries S. Identification of Plant Parts Using Color and Geometric Image Data.Translations of the ASAE.1993.36(5):1493-1500.
[8]Shimizu H, Heins R D. Computer vision based system for plant growth analysis. Trans of the ASAE.1995.38(3):959-964.
[9]林开颜,徐立鸿,吴军辉.计算机视觉技术在作物生长监测中的研究进展.农业工程学报.2004.20(4)：279-282.
[10]陈顺三.奶牛体形线性评定的图像处理技术研究.北京农业工程大学博士论文,1996.25-40
[11]张建平.数字图像处理在农产品质量检测中的应用一烟叶质量的自动分级.江苏理工大学博士论文,1994.55-65.
[12]刘禾.计算机视觉在水果自动分级中的应用.中国农业大学博士论文,1994.58-70
[13]宋韬.基于计算机图像处理的作物籽粒的识别研究.中国农业大学博士论文,1994.69-80
[14]杨秀坤,陈晓光,马成林等.用遗传神经网络方法进行苹果颜色自动检测的研究农业机械学报.1997.13(2)：76-173.
[15]陈佳娟,纪寿文等.基于遗传神经网络的玉米叶色的自动测定研究.农业工程学报.2000.15(5)：115-117.
[16]毛罕平,徐贵力,李萍萍.番茄缺素叶片的图像特征提取和优化选择研究.农业工程学报.2003.19(2)：133-136.
[17]张香琴,方如明,汪萍等.家蚕微粒子病的图像识别技术研究.农业机械学报.2001.32(5)：65-68
[18]李长缨,腾光辉,赵春江等.利用计算机视觉技术实现对温室植物生长的无损监测.农业工程学报.2003.19(3)：140-143.
[19]田有文,李成华.基于统计模式识别的植物病害彩色图像分割方法.吉林大学学报,2004.4,34(2)：291-293.
[20]马骏,王建华.一种基于数学形态学的植物病虫识别方法.深圳大学学报,2004,21(1)：72-75.
[21]Young J Han, Virgil L Quisenberry. Image Analysis to Quality Soil Pore Geometry. ASAE Paper, No.89-3004.
[22]Liao K, Paulsen M R, Reid J F. Real-time detection of color and surface defects of maize kernels using machine vision. Journal of Agriculture Engineering Research.1994,59:263～271.
[23]Shimizu H, Heins R D. Computer vision based system for plant growth analysis. Trans of the ASAE.1995,38(3):959～964.
[24]纪寿文,王荣本,陈佳娟等.应用计算机图像处理技术识别玉米田间杂草的研究[J].农业工程学报,2001,17(2)：154-156.
[25]吕俊伟,马成林,左春栓.用彩色图像识别田间洋葱和杂草的研究[J].农业工程学报,2001,17(5)：153-158.
[26]徐黎.VC++下的数字图像处理时限与研究[J].电脑开发与应用,2001,(4)：31-33.
[27]白耀辉,陈明,王举群.利用朴素贝叶斯方法实现异常检测.计算机工程与应用.352005,131-132.
[28]肖波,徐蔚然.基于贝叶斯分类器的混排文字切分与分类.计算机工程与应用.102005,83-85.
[29]吴立增,朱永利,苑津莎.基于贝叶斯网络分类器的变压器综合故障诊断方法.电工技术学报.20(4)2005,45-51.
[30]张少中,王秀坤,孙莹光.贝叶斯网络及其在决策支持系统中的应用.计算机工程.30(10)2004,1-3.
[31]陈晓光,于海业等.应用图像处理技术进行蔬菜苗特征量识别.农业工程学报.11(4)1995,23-26.
[32]安冈善文.图像处理技术在环境中的应用.电气学会杂志特集.1985,455-458.
[33]穗波雄信.根据图像提取植物的生长信息.农业机械学会关西支部第6次支部研究资料.10,1989,1-2.
[34]Bai Zhongying, Li Shaokun. The application of recognizing population feature of wheat images based on BP neural network. Proceedings of International Symposium on Future Software Technology, ISFST'2002. October 23-25,2002 231-236.
[35]Antonio J, et al. An evaluation of the application of infrared thermalimaging to the tele-diagnosis of sarcoptic mangein the Spanish ibex (Capra pyrenaica) Veterinary Parasitology. 109,2002,111-117.
[36]Cheng H D, et al. A novel fuzzy entropy approach to image enhancement and thresholding Signal Processing.75,1999,277-301.
[37]Cheng H D, et al. A Hierarchical Approach to Color Image Segmentation Using Homogeneity. IEEE Trans, on Image Processing.9(12)2000,2071-2082.
[38]Granitto, et al. Large scale investigation of weed seed identification by machine vision Computer and Electronics in Agriculture.2004.
[39]Ong S H, et al. Segmentation of color images using a two-stage self-organizing network Image and Vision Computing.20,2002,279-289.
[40]钱玲飞,刘玉树,李侃.朴素贝叶斯分类器在地形评估中的应用办法.计算机工程与应用.12.2005,189-191.
[41]王拓,蒋志雄,王颖初等.基于机器视觉的地膜下方棉苗的识别与定位.计算机工程与应用.24,2005：215-217
[42]钱玲飞,刘玉树,李侃.朴素贝叶斯分类器在地形评估中的应用办法.计算机工程与应用.12.2005,189-191.
[43]章毓晋.图象处理和分析.清华大学出版社.1999：219-220.
[44]李文化.模糊聚类新算法与模糊聚类神经网络.西安电子科技大学博士论文,1995.35-79.
[45]裴继红.基于模糊信息处理的图像分割方法研究.西安电子科技大学博士文,1998.50-89.
[46]吕铁英,彭嘉雄.图像轮廓特征提取新方法研究.中国图象图形学报.1999.4(8)：655-658.
[47]孙即样,王晓华,钟山等.模式识别中的特征提取与计算机视觉不变量.第一版.北京：国防工业出版社,2001.21-43.