Journal of Agricultural Science and Technology ›› 2025, Vol. 27 ›› Issue (1): 107-117.DOI: 10.13304/j.nykjdb.2023.0212

• INTELLIGENT AGRICULTURE & AGRICULTURAL MACHINERY • Previous Articles    

Research on Green Vegetation Extraction Method Based on Visible Light Band

Shuyuan ZHENG1(), Jian DAO1(), Xuelin ZHANG1, Shanshan LIU2(), Jianxiong WANG1()   

  1. 1.College of Water Conservancy,Yunnan Agricultural University,Kunming 650201,China
    2.College of Economics and Management,Southwest Forestry University,Kunming 650224,China
  • Received:2023-03-21 Accepted:2024-03-22 Online:2025-01-15 Published:2025-01-21
  • Contact: Shanshan LIU,Jianxiong WANG

基于可见光波段的绿色植被提取方法研究

郑舒元1(), 刀剑1(), 张学林1, 刘珊珊2(), 王建雄1()   

  1. 1.云南农业大学水利学院,昆明 650201
    2.西南林业大学经济管理学院,昆明 650224
  • 通讯作者: 刘珊珊,王建雄
  • 作者简介:郑舒元 E-mail:952012141@QQ.com
    刀剑 E-mail:daojian163@163.com第一联系人:郑舒元和刀剑为共同第一作者。
  • 基金资助:
    云南省教育厅基金项目(2024Y304);西南林业大学校级科研专项(110223010)

Abstract:

The visible light vegetation index is a method for vegetation extraction applied to RGB images. Various visible light vegetation indices have been utilized for vegetation extraction in unmanned aerial vehicle (UAV) images. However, it still need to improve in terms of extraction efficiency and accuracy. Based on the analysis of the spectral characteristics of green healthy vegetation and 8 different land cover types in UAV RGB images, this study proposed a visible light vegetation index based on the green and blue bands, named as enhanced green-blue ratio index (EGBRI). The effectiveness of EGBRI was compared with 8 commonly used visible light vegetation indices, and the accuracy was quantitatively evaluated by combining the results of visual interpretation of land cover and confusion matrix. The results indicated that the vegetation index calculated by EGBRI could effectively extract green vegetation in the study area and had a suppressive effect on other land cover types. Compared with other commonvisible light vegetation indices, EGBRI enhanced the discrimination between vegetation and other land cover types, demonstrating higher classification accuracy. The overall accuracy of EGBRI was 95.06%, the Kappa coefficient was 0.889 5, which was at a relatively high level, enabling rapid and accurate extraction of vegetation cover in the study area. In conclusion, the proposed EGBRI could effectively, rapidly, and accurately extract green vegetation information from UAV RGB images, demonstrating good applicability and extraction accuracy.

Key words: UAV, RGB shadow image, visible light vegetation index, green vegetation extraction

摘要:

可见光植被指数是一种基于RGB影像的植被提取方法,目前已被广泛用于无人机影像植被提取,现有的方法在植被提取效率及增大植被与其他地物区分度方面仍有可改进的空间。基于绿色健康植被光谱特性及8种不同地物在无人机RGB影像中的光谱特征,提出一种基于绿、蓝波段的可见光植被指数——超绿蓝比值指数(enhanced green blue ratio index,EGBRI),利用该指数与其他8种常见可见光植被指数提取效果进行对比研究,并采用基于目视解译的地物判别结果结合混淆矩阵进行精度量化评价。结果表明:由EGBRI计算的植被指数能够有效提取试验区绿色植被,对其他地物具有抑制作用;相比其他常见可见光植被指数,EGBRI增强了植被与其他地类的区分度,其分类精度更高,EGBRI总体精度为95.06%,Kappa系数为0.889 5,处于较高水平,能够对试验区的植被覆盖区域进行快速、准确的提取。研究结果表明,提出的超绿蓝比值指数(EGBRI)能够有效、快速、高精度、大范围地提取无人机影像RGB波段影像中绿色植被信息,且具有较好的适用性和提取精度。

关键词: 无人机, RGB影像, 可见光植被指数, 绿色植被提取

CLC Number: