欢迎访问《草业学报》官方网站,今天是 分享到:

草业学报 ›› 2014, Vol. 23 ›› Issue (6): 20-27.DOI: 10.11686/cyxb20140603

• 论文 • 上一篇    下一篇

基于数码相机的草地植被盖度测量方法对比研究

陈祖刚1,2,巴图娜存1,徐芝英1,2,胡云锋1,*   

  1. 1.中国科学院地理科学与资源研究所,北京 100101;
    2.中国科学院大学,北京 100049
  • 收稿日期:2013-11-11 出版日期:2014-12-20 发布日期:2014-12-20
  • 通讯作者: E-mail:huyf@lreis.ac.cn
  • 作者简介:陈祖刚(1989-),男,河南信阳人,在读硕士
  • 基金资助:
    国家973计划(2010CB950904),国家自然科学基金(40971223)和中国科学院知识创新方向性项目(KZCX2-EW-306)资助

Measuring grassland vegetation cover using digital camera images

CHEN Zu-gang1,2,BATU Nacun1,XU Zhi-ying1,2,HU Yun-feng1   

  1. 1.Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China;
    2.University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2013-11-11 Online:2014-12-20 Published:2014-12-20

摘要: 草地植被盖度是表征生态系统植被生长状况及环境质量的重要参数。在草地植物群落野外调查中,可以利用数码相机拍摄草地样方照片,而后在室内利用图像处理软件进行自动或半自动的植被盖度测量。随着移动智能设备(如iPhone/iPAD或各类Android Phone/PAD)的快速发展和普及,野外实时获取草地样方照片,同步计算草地植被盖度,并与有关遥感反演参数产品作校验对比分析,将成为未来地学移动测量和研究的重要方向。本研究在总结梳理既有利用数码相机识别植被盖度方法的基础上,设计了低覆盖、中低覆盖、中等覆盖、中高覆盖、高覆盖5种不同植被盖度情景,以及从早上6:00 到下午6:00、每隔2 h一次、全天共7次不同光照环境下的照相方案。继而以Photoshop人工勾勒和测算方法为基准,选择RGB阈值法、RGB决策树法、HSV判别法3种自动测量方法开展对比研究。测量结果的对比分析表明,草地盖度变化对RGB阈值法和HSV判别法的盖度识别精度无明显规律性影响,RGB决策树的盖度识别精度随着草地盖度的增加而增加;光照强度越强,RGB阈值法和HSV判别法对同一草地样方估算的盖度值越小,RGB决策树法估算的盖度值随光照强度的变化没有固定的规律。总体上讲,RGB阈值法和HSV判别法的识别精度较高,RGB决策树法误判率较高,但后者可以识别出非绿色的植物茎、花朵。最后提出了在现有绿色植被像元识别方法的基础上,结合边缘检查算法等图形像素的统计学特征分析方法,能进一步提高草地植被盖度测量的准确率。

Abstract: Grassland vegetation coverage is an important parameter for characterizing vegetation condition in ecological research. With the popularization of digital devices, such as digital cameras, vegetation coverage can be measured in automatic or semi-automatic ways by analyzing the digital photographs using image processing platforms. Moreover, the rapid development and popularity of smart mobile devices (such as the iPhone/iPad) means that there is an opportunity to investigate the use of these devices for real-time collection, processing and analyzing of data such as vegetation cover. Five quadrats with different vegetation cover were selected and measured in different lighting conditions. Photographs were taken every 2 h from 6 AM to 6 PM. The RGB threshold value method, RGB decision tree method and HSV discriminant method were selected for calculating vegetation cover. When light intensity was constant, vegetation cover had no effect on the measurement precision of the RGB threshold value method and the HSV discriminant method, but greatly influenced the measurement precision of the RGB decision tree method. The RGB threshold value method and the HSV discriminant method were influenced by light condition; when light intensity increased cover estimates declined. However, there did not appear to be a good relationship between light intensity and cover estimates for the RGB threshold value method. Overall, the measurement precision of the RGB threshold value and HSV discriminant methods were higher than the RGB decision tree method but the latter could identify plant stems and flowers that were not green. Using the current green plant pixel identification methods combined with improved statistical analysis methods such as algorithms able to examine edge pixels further improvement of the precision of this technique could be achieved.

中图分类号: