欢迎访问林业科学,今天是

林业科学 ›› 2013, Vol. 49 ›› Issue (10): 80-87.doi: 10.11707/j.1001-7488.20131013

• 论文与研究报告 • 上一篇    下一篇

面向对象多尺度分割的SPOT5影像毛竹林专题信息提取

孙晓艳1,2, 杜华强1,2, 韩凝1,2, 葛宏立1,2, 谷成燕1,2   

  1. 1. 浙江省森林生态系统碳循环与固碳减排重点实验室 浙江农林大学 临安 311300;
    2. 浙江农林大学环境与资源学院 临安 311300
  • 收稿日期:2012-12-12 修回日期:2013-04-02 出版日期:2013-10-25 发布日期:2013-11-05
  • 通讯作者: 韩凝
  • 基金资助:

    国家自然科学基金项目(31070564);浙江省林业碳汇与计量创新团队项目(2012R10030-01);浙江省自然科学基金项目(Y3100427;LQ13C160002);浙江农林大学科研发展基金(2012FR073)。

Multi-Scale Segmentation, Object-Based Extraction of Moso Bamboo Forest from SPOT 5 Imagery

Sun Xiaoyan1,2, Du Huaqiang1,2, Han Ning1,2, Ge Hongli1,2, Gu Chengyan1,2   

  1. 1. Zhejiang Provincial Key Laboratory of Carbon Cycling in Forest Ecosystems and Carbon Sequestration Zhejiang A & F University Lin'an 311300;
    2. School of Environmental and Resources Science, Zhejiang A & F University Lin'an 311300
  • Received:2012-12-12 Revised:2013-04-02 Online:2013-10-25 Published:2013-11-05

摘要:

以SPOT5卫星遥感影像为基础,采用面向对象的多尺度分割方法,建立类层次结构,提取毛竹林遥感专题信息。结果表明: 1) 对毛竹林而言,在SPOT5红、绿、蓝3个波段上的最佳纹理窗口大小分别为9×9,7×7,9×9,比较接近; 2) 面向对象的多尺度分割方法能较为精确地提取毛竹林专题信息,用户精度达到90%以上,高于最大似然法提取毛竹林的信息精度(88.57%); 3) 增加纹理信息的多尺度分割方案既保证了毛竹林专题信息的提取精度(92.16%),又兼顾了其他森林类型,分类总精度、Kappa分别为92%和88.14%,为本研究的最高精度。

关键词: 面向对象, 多尺度分割, 毛竹林, 信息提取, SPOT5

Abstract:

Based on SPOT5 remotely sensed imagery, this research focused on delineating moso bamboo forest using object-based method, which provided the advantages of multi-scale segmentation and developing hierarchical structure. The results showed that: 1) The most appropriate window sizes for calculating texture using red (R), green (G) and blue (B) band in SPOT5 image were 9×9,7×7,9×9; 2) Extracting moso bamboo using multi-scale segmentation technique of object-based method was more accurate, with the producer's accuracy reaching 90%, obviously higher than that of the conventional maximum likelihood method(88.57%); 3) Multiresolution segmentation with the aid of texture not only ensured the accuracy of moso bamboo, but also provided help to the other forest types. The overall accuracy was 92% and the Kappa coefficient was 88.14%, both of which were the highest accuracy in the present study.

Key words: object-oriented, multi-scale segmentation, moso bamboo forest, information extraction, SPOT5

中图分类号: