分类号：密级：
智能婴儿床UDC ：编号：
识别系统研究
The Tobacco Plant Diseases and Insect Pests Image Intelligent Recognition System Based On Android Mobile Terminal
硕士研究生：吴子龙
青梅1H指导教师：杨毅教授
申请学位类别：农业推广硕士
领域：农业信息化
培养学院：基础与信息工程学院
huae
二O一五年五月
独创性声明
本人声明所呈交的学位论文是本人在导师指导下进行的研究工作及取得的研究成果。尽我所知，除了文中特别加以标注和致谢的地方外，论文中不包含其他人已经发表或撰写过的研究成果，也不包含为获得云南农业大学或其它教育机构的学位或证书而使用过的材料。与我一同工作的同志对本研究所做的任何贡献均已在论文中作了明确的说明并表示了谢意。 研究生签名：时间：年月日
关于学位论文使用授权的声明本人完全了解云南农业大学有关保存、使用学位论文的管理办法及规定，即：云南农业大学有权保留送交论文的复印件和磁盘，允许论文被查阅和借阅，可以采用影印、缩印或扫描等复制手段保存、汇编学位论文。同意云南农业大学可以用不同方式在不同媒体上发表、传播学位论文的全部或部分内容。
(保密的学位论文在解密后应遵守此协议)
研究生签名：时间：年月日
导师签名：时间：年月日
摘要
烟草是我国农业种植中重要的经济作物，在2013年云南省烟草总产量占我国烟草总产量的31.89%，可以看出云南省的烟草产量的会对我国烟草供给产生非常大的影响。其中烟草病虫害的防治是影响烟草的产量重要因素之一。因此在云南省烟草病虫害的预防、诊断、技术显得尤为重要。论文以烟草赤星病、病、角斑病以及斜纹夜蛾等烟草病虫害为例，进行图像识别研究。
本文以提高烟草农户的烟草病虫害诊断效率以及诊断水平为目的，做了以下工作。首先，研究了国内外在农业病害虫图像识别领域发展状况并进行了总结，提出了基于Android移动终端烟草病虫害图像智能识别系统。系统在Android平台上综合了数字图像处理技术、网络通信、数据库、植物病理学以及模式识别等知识进行烟草病虫害图像智能识别研究，系统采用C/S架构，包括Android客户端、服务器端、PC端。
Android移动终端：烟草农户进入Android客户端后要添加作物病斑图片，要添加的图片可以从移动终
端相册添加也可以通过移动终端自带的相机拍摄田间图片。在添加照片成功后，Android客户端会对图片进行预处理、图像分割、特征提取。然后，Android移动终端通过TCP/IP协议将处理得到特征值发送到服务器端。最后，Android移动终端会得到服务器端返回的检测结果。
飞轮壳服务器端：服务器端一直处于监听是否有消息的状态，一旦接收到Android移动终端发送过来的特征值，服务器就会自动的通过该特征值调用图像相似度算法进行检索特征数据库。最后将检索后的结果返回给Android移动终端。
冯代存PC端：植物病虫害诊断人员可在PC客户端提取烟草病虫害图像特征，并将提取的图像特征存储到烟草病虫害图像特征数据库。
本系统的特：第一，烟草农户可以通过Android智能终端实现拍摄烟草病虫害的照片或者在终端相册中添加烟草病虫照片；第二，Android客户端可对病虫害图像进行分割、提取特征；第三，实现了Android客户端与服务器的通信，联网识别病害种类及诊断方法。
关键词：图像识别；烟草病虫害；Android；图像分割；图像特征提取
The Tobacco Plant Diseases and Insect Pests Image Intelligent Recognition System Based On Android Mobile Terminal
Wu Zilong
Directed by Yang Yi
Abstract
Tobacco is an important economic crop in agriculture in our country. In 2013, the production of tobacco in Yunnan Province accounted for 31.89% of the total production in China. It can be seen that in Yunnan province tobacco production can produce a very big impact to our country tobacco supply. The prevention and control of tobacco plant diseases and insect pests is one of the most important factors of affecting the output of tobacco. Therefore, the prevention, diagnosis and treatment technology of tobacco plant diseases and insect pests are particularly important in Yunnan.
In this paper, in line with the raise the level of tobacco pests and diseases diagnosis. First of all, the paper has studied and summarized the development situation of domestic and external in agricultural pest image recognition. Then the paper has established a tobacco plant disease image intelligent identification system based on the Android mobile terminal, which uses C/S architecture, including mobile client and the server.
Mobile client: first of all, after entering the client to add crop disease spot image, which can be added from the phone album or take a picture using the phone camera. Secondly, after adding photos and submitting, the mobile phone will process the image with image preprocessing, image segmentation and feature extraction. Then, mobile terminal will process the obtained eigenvalue via TCP/IP communication protocol and send it to server-side. Finally, the mobile client will receive the results returned by the
server.
Server: firstly, the server system has being in monitoring the state of the message is received or not. Once the server receives the eigenvalues from the mobile phone, the server will automatically call image similarity algorithm by the feature value and then search retrieval lesion characteristics database. Finally, the retrieval result will be returned to the mobile client.
海洋浮标PC Client: plant diseases and insect pests diagnosis personnel can be extracted in the PC client tobacco plant diseases and insect pests image characteristics, and the extracted image features stored in the tobacco plant diseases and insect pests image database.
The characteristics of this system: first, the farmers can add crop disease spot image, which can be
added from the phone album or take a picture using the phone camera. Second, the Android client can process the image with image segmentation and feature extraction; Third, the paper implements the communication between Android client and server, the identification of diseases with network and the diagnostic method.
Key Words:Image recognition; The tobacco plant diseases and insect pests; Android; Image segmentation; Image feature extraction
豫公网安备41160202000603
站长QQ:729038198
关于我们
投诉建议