【作者】 刘九庆；

【导师】 李东升； 汤晓华；

【作者基本信息】 东北林业大学 ， 机械设计及理论， 2005， 博士

【摘要】 根系是植物的重要组成部分,是植物生长发育、新陈代谢的主要营养器官。根系形态参数是反映根系生长状况的主要因素,通过定期测量和分析植物根系形态参数对于了解、研究植物的生长发育状况以及防治病虫灾害均具有重要意义。但由于根系生长在地下,准确取样、测定、观察存在一定困难,而目前现有的挖掘法、整段标本法、剖面法等采样方法,都有采样破坏性和工作量大等缺点,因而阻碍了根系研究的深入开展。 本文是基于计算机视觉技术、机电一体化技术和计算机图像处理技术,研究开发了植物根系图像监测分析系统。系统以CCD摄像机、图像采集卡、计算机、三自由度根系图像采集运动系统以及根系形态参数分析软件为主体构建而成,适于植物根系形态监测及其采集分析。实现了对根系大视场观测和形态参数的快速准确测量与分析,为研究植物根系提供一种全新、直观、高效、可靠的研究手段。本文对CCD摄像机标定、图像几何畸变校正、图像拼接、图像边缘检测等方面进行了细致深入的研究,提出了一种基于分形理论H参数计算的改进方法——增容矩阵算法,用以弥补传统算法图像边缘缺损的缺陷进而提取植物根系轮廓,在后续的根系图像分析实验中取得了很好的效果。 本论文的主要研究工作包括: 1、植物根系图像监测分析系统的研究设计。硬件包括:CCD摄像机三自由度精确定位机构的研究设计,电气控制系统的设计,照明系统以及图像采集传输系统的设计等;软件包括根系图像采集,根系图像处理和根系形态参数分析三部分组成。 2、运用数字图像处理技术对所采集的根系数字图像进行了研究,包括根系图像几何畸变校正、根系图像拼接、根系图像边缘检测及根系图像形态参数计算等方面。 3、提出了运用分形理论提取植物根系轮廓的方法,并和传统的Sobel算子、Canny算子进行比较,实验结果表明分形算法在根系图像边缘提取和噪声抗干扰能力方面优于其它算子。在H参数值的计算上本文提出了一种改进的增容矩阵算法,给出了增容矩阵填补规则,实验证明该算法在根系图像边缘检测中的可行性。 基于计算机视觉技术全自动/手动双功能植物根系全景图像采集处理分析系统(NEFU-PRSAS-01)的研究突破了国内在植物根系无损在线监测多项瓶颈。在根系图像摄取范围、操作的灵活性、自动化程度等方面均取得了突破。特别是首次将分形理论应用于植物根系图像轮廓提取。在此基础上开发的软件系统(NEFU-PRSAS-S01)实现了植物根系形态参数自动分析计算。为我国农林业相关领域的科研提供了一种全新的研究手段,为推进我国植物根系研究进行了一次有益的尝试。 本文是在哈尔滨市青年研究基金“植物根系图像监测分析系统的研究”课题资助下研究完成的。更多还原

【Abstract】 Root is one of the most important components of plant, also is the main alimentative organ of growth and metabolism. Root conformational parameters is one of the main factors of reflecting plant growth, so it is very useful to find out plant growth, prevent plant from insects and other applications by measuring and studying root conformational parameters periodically. But underground root is hard to sample, measure and observe accurately. Especially the existent methods such as Digging, Whole segment sample, Section and so fothr are time consuming and are devastating, which will block the deep study of root.This thesis combines the technologies computer vision analysis, mechtronics and image processing to exploit a plant root analysis system. The system consists of a CCD vidicon, an image gathering card, a three-dimensional motion image gathering components and analyzing software of conformational parameters. It is suitable for measurement of root conformational parameters and image monitoring and analyzing. This will make the fast and precise measurement and analysis of root conformational parameters possible, also will provide a fire-new, high effective and reliable tool to study the roots of plants. We have already gone for deep researches in vidicon demarcation, emendation of image geometric aberrance, images joint, edge detection of plants’ roots basing on fractal theory and other aspects. We also posed an improved method of H parameter computation—incremental matrix algorithm to pick-up the outline of plants’ roots. This will handle the shortcomes of traditional methods and gained much better experimental results afterwards. The main researching work of this thesis:1、 To design the system monitoring and analyzing image of plant root. The hardware consists of the three-dimensional precisely orientating machine, the electrical control system, the lighting system and so on. The software includes the control of vidicon’s orientating electromotor, the emendation and mosaic of root image, and the analysis of root conformational parameters.2、 To study the processing of root image deeply using digital image processing technique. That includes emendation of image geometric aberrance of short focus lens, mosaic of root image’s panorama, calculation of root conformational parameters and others. 3、 Our research takes advantage of the fractal parameter to pick-up roots’ image, and contrasts with the traditional Sobel Operator and Canny Operator. The outcome indicates that method using fractal parameter excels others in picking-up roots’ image. In the light of calculation of parameter H, we mention an improved method called incremental matrix which is founded on the original arithmetic. We also make rules for the incremental matrix. Experiments indicate that this method is feasible in picking-up roots’ image and calculation of parameter H.The exploit of NEFU-PRSA-01 auto/manual double-founction plant root panorama imageprocessing and analyzing system, which based on computer vidicon, is a breakthrough around the country in many choke points of online underground root monitoring. In the lights image gathering range, operational flexibility, automatic extent and other aspects, we have all obtained some breakthrough. Especially that the system applies fractal theory on picking-up roots’ image for the first time. Besides these, we have exploited the software NEFU-PRSA-S01 used to analyse conformational parameters of root automatically. This will provide a fire-new studying method for scientific research related to agriculture and forest. And it is all a try that will advance the research of root in our country.This thesis is supported by the project Research on plant root image monitoing and analyzing system of the Harbin Youth Studying Foundation.更多还原