CAJ | 학술논문

对福建省将乐国有林场地物进行分析，结合QuickBird高分辨率影像及其重采样低分辨率影像进行多层次分割，同时结合ESP分割尺度评价工具获取最优分割尺度参数，以实现影像地物最佳分割效果。对不同分割层次下对象光谱特征、纹理特征等进行提取，应用专家知识及统计特征选取影像对象特征和确定阈值，实现研究区域森林地物及树种的分类。对森林植被等分类总精度达到80．53％。
대복건성장악국유림장지물진행분석，결합QuickBird고분변솔영상급기중채양저분변솔영상진행다층차분할，동시결합ESP분할척도평개공구획취최우분할척도삼수，이실현영상지물최가분할효과。대불동분할층차하대상광보특정、문리특정등진행제취，응용전가지식급통계특정선취영상대상특정화학정역치，실현연구구역삼임지물급수충적분류。대삼림식피등분류총정도체도80．53％。
An investigation was conducted by analyzing the groundcover of Jiangle Forest Farm in Fujian Province, and multi-level segmentation was conducted based on QuickBird high spatial resolution image and its resampling multi-resolution ima-ges by using the ESP ( Estimating the Scale Parameter) to acquire optimal segmentation parameters for different groundcov-ers.The segmentations achieved good results.With the extraction of spectral, textural features, and confirming the thresh-old by expert knowledge and statistical features, the forest groundcover classification was accomplished with the total accu-racy of 80.53%.