毕业论文-潘伟-修订一版.docxVIP

Abstract 石河子大学 信息科学与技术学院毕业论文 课题名称：基于机器视觉的棉花叶面蚜虫计数算法研究学生姓名：潘伟学 号：2011508175学 院：信息科学与技术学院专业年级：电子信息工程2011级（1）班指导教师：查志华职 称：讲师完成日期：二○一五年六月三日 摘要  PAGE \* MERGEFORMAT IV  PAGE \* MERGEFORMAT III 基于机器视觉的棉花叶面蚜虫计数算法研究 学生 ：潘伟 指导老师 ：查志华 [摘要] 作物上害虫种群密度和危害程度是害虫防治决策的重要依据，也是精确喷药的关键信息。与人工测虫方法相比，使用计算机视觉来自动获取害虫信息，不仅可降低劳动强度、提高工作效率，而且便于与后续的防治决策和精确施药实现技术对接和技术集成。当前，害虫检测和计数的主要困难之一还是小虫体的检测和计数。就小虫体的计算机视觉检测和计数而言，也存在着因虫体几何尺寸小而导致图像处理的困难。其一，在同样的成像设备和成像条件下，小虫体的成像质量一般比大虫体要差，这对小虫体图像特征提取、处理和检测都有一些新要求。其二，小虫体构成的虫群图像粘连很严重，这对自动化快速准确计数带来了困难。为此，本文以棉花叶面蚜虫为对象，研究基于图像信息的蚜虫检测和计数方法。 [关键词] 蚜虫,图像处理,k-means聚类算法,阈值分割,腐蚀重建,计数Abstract Abstract Study on the counting algorithm of cotton leaf aphids based on machine vision Students：Pan Wei Teacher：ZhaZhihua Abstract: Crop pest population density and harmful levels are important criteria for pest control，decision making，and the key information for precise spraying．Compared with manual measurement insects，using computer vision to automatically obtain the pest information，not only can reduce labor intensity, improve efficiency, and easier to control and decision—making with precise spraying realize technology docking and integration．Currently, one of the main difficulties for the pest detection and counting is small worm detection and counting．In terms of the computer vision on a small worm detection and counting，the small worm also exist the difficulty of image processing what be caused by small geometry．First，in the same imaging equipment and imaging conditions，image quality of small worm is worse than large worm generally, this brings new request to small worm that image feature extraction，processing and detection．Second，The image overlapping of insect group which constitute with the small worm is very serious，this brings difficult to automation，rapid and accurate counting．So this paper targeted to the cotton leaf aphids，research on the image detection and counting methods for aphids based on the image information. Key w