计算机应用 ›› 2010, Vol. 30 ›› Issue (12): 3269-3273.

• 图形图像处理 • 上一篇    下一篇

遥感图像林区自动阈值分割算法及性能比较

崔天意1,刘文萍2,张宁3   

  1. 1. 北京林业大学 信息学院
    2. 北京林业大学
    3.
  • 收稿日期:2010-06-04 修回日期:2010-07-21 发布日期:2010-12-22 出版日期:2010-12-01
  • 通讯作者: 崔天意
  • 基金资助:
    国家自然科学基金资助项目;国家重点基础研究发展(973)计划资助项目;北京林业大学科技创新计划资助项目

Algorithms and performance comparison of automatic thresholding segmentation for forest regions in remote sensing image

  • Received:2010-06-04 Revised:2010-07-21 Online:2010-12-22 Published:2010-12-01

摘要: 选取了几种经典的自动阈值选取算法对高分辨率遥感图像林区目标进行分割实验,并引入错分类误差、形状测度、均匀测度、最终测量精度和运算速度作为算法评判准则,客观、定量地比较了各种算法对高分辨率遥感图像林区目标的分割性能,所得结论对林区目标分割方法的选取具有一定的指导作用。

关键词: 高分辨率遥感图像, 图像分割, 自动阈值选取, 评判准则, 林区

Abstract: Several classic automatic threshold-selecting algorithms were selected for segmenting forest regions in high-resolution remote sensing image. As evaluation criteria of algorithms, misclassification error, shape measure, uniformity measure, relative ultimate measurement accuracy and running time were implemented to compare segmentation performance of each algorithm objectively and quantitatively. The conclusion can play a guidance role in automatic threshold-selecting methods to extract forest regions in high-resolution remote sensing image.

Key words: high-resolution remote sensing image, image segmentation, automatic threshold-selecting, evaluation criterion, forest region