【作者】 胡晓荣；

【导师】 周春国；

【作者基本信息】 南京林业大学 ， 森林经理学， 2007， 硕士

【摘要】 从树木生长年轮的变化中,我们可以了解树木历年的生长、培育情况,并且通过对树木年轮的研究,有助于我们了解当时当地气候变化、环境污染等情况。在以往的研究中,对树木年轮的研究主要集中在对年轮宽度、年轮密度等几个指标的量测与分析上。传统的年轮分析方法都是依据人眼对年轮进行识别,再通过量测工具获取年轮宽度数据。这种依靠人工操作读取年轮宽度的方法不仅效率低,而且工作繁琐、费时费力。更重要的是在判读年轮、量测年轮宽度时,由于年轮边缘延伸的不规则变化可能会给测量结果带来随机波动的影响,而且人眼的观测误差和疲劳错误会进一步降低实际测量的精度。上面这些因素都成为了将大量采集样本转换成分析数据的瓶颈。近几年,随着计算机技术的飞跃发展,信息化技术正在改变着传统林业的研究方式,特别是计算机图像处理技术的发展和应用,使得设计出一套自动识别定标树木年轮的系统成为了可能。目前世界上较成熟的年轮分析系统主要是加拿大REGENT公司的WinDENDRO和德国Frank Rinn公司的LINTAB,本文的研究正是基于WinDENDRO年轮分析系统对多种类型的年轮图像进行了自动识别测量。研究实验表明,系统所提供的年轮分析功能对于理想图像样本识别效率高,而实际应用中年轮图像可能存在着色调变化不均匀、树脂道斑、打磨痕迹等干扰信息,系统识别的效果很不理想。针对以上情况,本文探讨了利用计算机图像处理的几种方法,对实际采集的年轮图像进行直方图修改、平滑滤波、二值化处理等处理,使处理后的图像在应用到WinDENDRO分析系统时,能快速准确的计算出年轮个数、宽度值等参数,为后续树木生长分析提供精确的数据,解决人工操作费时费力和错漏等问题。同时,本文还探讨了利用图像分割、数学形态学等方法,对树木年轮进行边缘提取。通过本文所作的研究,希望能推动树木年轮分析方法的进步,为数字化林业提供高效、精确的数据源。更多还原

【Abstract】 From the change of the arboreal growth annual ring, we may understand trees all previous years the growth, the cultivation situation, and with the research on tree ring, we get the situation of climatic change and environmental pollution.Presently research mainly concentrates on the tree ring in the annual ring width, the annual ring density, etc. Traditional annual ring analysis method rests on the human eye to the annual ring recognition, with the aid of in measuring tools to get annual ring width data. This method brings low efficiency, processing method tedious, time-consuming and hard sledding. And when interpret annual ring, gauging annual ring width, because the annual ring edge extends each kind of irregular change brings the stochastic undulation influence for the measuring technique, and the human eye observation error and weary wrong also can reduce the actual measuring accuracy greatly, all of which restricted the development of fast transformation analysis data on massive sample with high effective.With the leap development computer processing technology, info-tech is changing the traditional forestry research method, which makes it possible to design a set of automatic recognition calibration system on tree ring. At present, the most advanced annual ring analysis systems include Canada’s LINTAB tree ring analysis system and WinDENDRO annual ring analysis system, etc.Based on WinDENDRO annual ring analyzer, and carries on automatic recognition survey to several kind of type annual ring images. The research experiment indicated, the system provides the annual ring analysis function is high regarding the ideal image sample recognition efficiency, but in practical application, because the annual ring image originates various, in the annual ring image possibly has the tone to change, resin passage spot, polish trace and so on disturbance information’s non-uniform, the system recognition effect is not very ideal.In view of above condition, this article carries on the smooth noise reduction to the actual gathered annual ring image with the use of computer imagery processing method. With the application of WinDENDRO annual ring analysis system, users can quickly get the accurate data of computation annual ring number and the width value, which will provide accurate data for the following arboreal growth analysis, and to solve the problem of manual control time-consuming hard sledding and errors and omissions. Meanwhile, through methods of image division, mathematics morphology, this article also attempts to carry on the automatic recognition and the extraction to the annual ring edge, which will certainly improve the tree ring analysis method, to provide high effective and precise data pool for the digitized forestry.更多还原