CAJ | 학술논문

面向对象的影像分析已成为针对高分辨率遥感影像信息提取和行业应用的一种先进方法.该方法的实用效果,基础和关键在于对影像中所关注地物目标的影像分割提取效果.针对高分辨率遥感影像地面构筑物的提取,采用经验性符合统计比较法,把自动分割结果与人工分割提取的地面构筑物区域轮廓相对符合程度分为3个等级:良好、可以接受和不可接受,对当前应用广泛的两种多尺度影像分割方法进行了地面构筑物分割提取效果评估.
면향대상적영상분석이성위침대고분변솔요감영상신식제취화행업응용적일충선진방법.해방법적실용효과,기출화관건재우대영상중소관주지물목표적영상분할제취효과.침대고분변솔요감영상지면구축물적제취,채용경험성부합통계비교법,파자동분할결과여인공분할제취적지면구축물구역륜곽상대부합정도분위3개등급:량호、가이접수화불가접수,대당전응용엄범적량충다척도영상분할방법진행료지면구축물분할제취효과평고.
The preliminary and crucial step in object-based image analysis is the segmentation of the image. Therefore, the qualitative and quantitative assessment of segmentation results is very important for further image analysis as well as for choosing the appropriate approach for a given segmentation task. In this paper, an empirical discrepancy method using the relative ultimate correspondence statistical comparison, which reflects the similarity between the segmentation output and manually created reference building regions, has been applied to compare the segmentation results for building extraction. The relative ultimate correspondence is classified into three grades: Good, Acceptable and Unacceptable. The results of statistical comparison correspond with the ratings of building extraction accuracy.