Comparison of Stand-Level Delineation Algorithms Using Lidar Data

• Main Authors: CHANG,YU-SHENG, 張育昇
• Other Authors: CHANG,KUAN-TSUNG
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/22yyj6

碩士 === 明新科技大學 === 土木工程與環境資源管理系碩士班 === 106 === In recent years, carbon emissions and land use change are important issues that reduce global warming. The LiDAR technique is quite often used to estimate forest canopy parameters based on the generated canopy height model (CHM), and it is also very useful for estimating forest canopy density. In this study, the digital elevation model (DEM) is subtracted from the traditional algorithm digital surface model (DSM) and other CHMs can be generated by a pit-free process. Then, two algorithms, a multi-layer morphological active contour (MMAC) and a spatial wavelet analysis (SWA) were implemented in this study for individual tree rendering. Finally, the experimental results of the two stand-level estimation methods can be evaluated with manually measured canopy parameters (ie. tree height and crown diameter). The experimental results show that the extraction correctness is 89% with the SWA method using scale1-10 m in 0.1 increments. However, the MMAC method has a significant improvement with a 93% correct rate. Moreover, the success rate for two stand-level delineation algorithms using a pit-free CHM will be better than ones using a simple subtraction CHM