树木年轮研究的扫描图像分析方法及其在环境演变中的应用

METHOD OF SCANNING IMAGE ANALYSIS FOR TREE-RING RESEARCH AND ITS APPLICATION IN ENVIRONMENTAL CHANGE

本文主要探讨用普通扫描仪作为获取树轮信息手段的有效途径，以及研究将图像分析和数字处理技术应用于树木年轮学的方法。通过对树轮图像的分布特征和噪声变化规律的统计分析，本文设计出一套图像分割、信息提取算法和相应的处理程序。从实际的应用效果以及与其它树轮研究方法所作的对比分析结果表明，这种扫描图像处理方法不仅可行，而且高效、准确，拓宽了树木年轮学方法在环境演变研究中的应用途径。

The primary objective of this paper is to probe an effective way of acquiring thetree-ring view singular value with a plate scanner and of applying the image analysisand digital process to dendrochronological research. Through the stahstical analysis oftree-ring image distribution charateristics and noise variation, a new algorithm ofimage segmentation and feature extraction is designed, and then a correspondingsoftware package of personal computer tree-ring image process system is developedaccording to the dendrochronological research character. As a result of the realprocess and the analysis combined with other methods of dendrochronological research,it is demonstrated that the approach of scanning image process is not only feasiblebut also more effective and accurate. Therefore, it will be possible to make thedendrochronological research and applicahon in environmenta change more widespreadand profound. The contents and achievements of the research are mainly reflected inthe following six aspects:(1) Establishing an effechve method by using the personal computer devices(including tLhe plate scanner) to acquire the tree-ring view singular value, whichimproves the measuring tree-ring way. Compared with other specialized devices, thismethod reduces the processing cost and makes the processing more efflciency andmore accuracy.(2) Making a systematic study and analysis in the tree-ring and its errordistribution with the image theory and the mathematics method, and proving theresearch way to be feasible through the theory and the practice.(3) Making full use of the two-dimension information of digital image to studythe variation in the tree-ring growth area and then to raise the reliability of statisticalanalysis.(4) Thviding a ring into earlywood and latewood areas by the image segmentprocess so that the time-scale analysis precision will be improved anddendrochronological research will be more profound.(5) Designing a quantitative index of the difference between earlywood mean grayvalue and latewood mean gray value as the indicator of tree-ring growth. The indexcan effectively restrain the influence of image noise on the statistic analysis oftree-ring gray value, and have a good response to climate changes, which can be used as a subshtute for maximum density of tree-ring in dendrochronological research.(6) Developing a software package to be applied to dendrochronological analysisand research based on the personal computer using Microsoft Visual C+ + language.