内蒙古典型草原布氏田鼠鼠洞无人机遥感识别研究
刘楠 , 郭雪佳 , 褚东阳 , 林忆馨 , 李鑫 , 杨文蕾 , 叶国辉 , 付和平 , 袁帅
内蒙古农业大学学报(自然科学版) ›› 2026, Vol. 47 ›› Issue (5) : 36 -45.
内蒙古典型草原布氏田鼠鼠洞无人机遥感识别研究
UAV Remote Sensing Identification of Brandt's Vole Burrows in Typical Grasslands of Inner Mongolia
本研究旨在提升布氏田鼠(Lasiopodomys brandtii)在内蒙古典型草原上的监测效率,以便更有效地应对其挖掘活动导致的草场退化。传统的监测方法如堵洞盗洞法费时费力,已难以满足实际生产需求。针对此情况,本研究提出基于无人机遥感技术的布氏田鼠鼠洞监测方法,并于2023年在内蒙古锡林郭勒盟典型草原采集无人机多光谱影像数据,结合面向对象的多尺度分割、植被指数,以及支持向量机和K邻近值两种监督分类器对布氏田鼠鼠洞进行自动识别。在使用大疆M300 RTK无人机搭载RedEdge双相机成像系统的条件下(分辨率1 280×960像素),研究结果确定布氏田鼠鼠洞的最佳分割尺度为30 cm(波段权重比1∶1∶1∶1∶1,形状因子0.3,紧致因子0.5),并基于OSAVI的支持向量机(SVM)分类器表现最佳,总体精度达74.29%(Kappa系数0.719 3),显著优于K邻近值分类器(KNN)(总体精度72.14%,Kappa系数0.689 6,P<0.001)。本研究明确了布氏田鼠鼠洞识别的关键参数和最佳监督分类方法,为无人机在草原鼠害监测中的高效应用提供了可靠方法。
This study aimed to improve the monitoring efficiency of Brandt's vole (Lasiopodomys brandtii) in the typical grasslands of Inner Mongolia, and to more effectively address the grassland degradation caused by its burrowing activity. Traditional monitoring methods, such as the burrow-plugging method, are time-consuming and labor-intensive, failing to meet practical needs. In response, this study proposed a UAV remote sensing-based method for monitoring the Brandt's vole burrows. In 2023, multispectral imagery was collected via UAV in the typical grasslands of Xilin Gol, Inner Mongolia, and analyzed using object-oriented multi-scale segmentation, vegetation indices, and two supervised classifiers, i.e., support vector machine (SVM) and K-nearest neighbor (KNN), to automatically identify vole burrows. Under the conditions of using a DJI M300 RTK drone equipped with a RedEdge dual-camera imaging system(Resolution 1 280×960 pixels), the study determined that the optimal segmentation scale for Brandt's vole burrows was 30 cm, with the optimized soil-adjusted vegetation index (OSAVI) identified as the best vegetation index. SVM classification based on OSAVI yielded the highest recognition accuracy at 74.29%, with a Kappa coefficient of 0.719 3, outperforming KNN. This study clarified the key parameters and optimal supervised classification method for the vole burrow identification, providing a reliable approach for an efficient UAV-based pest monitoring method in grasslands.
| [1] |
付和平,卜凡,袁帅, |
| [2] |
付和平 . 草原主要鼠害[J]. 草原与草业,2020,32(3):63. |
| [3] |
卜凡,曹晓娟,张卓然, |
| [4] |
武晓东,曹润波,邹元平, |
| [5] |
|
| [6] |
徐正刚,赵运林,李波, |
| [7] |
|
| [8] |
李培先,郑江华,倪亦非, |
| [9] |
崔博超,郑江华,刘忠军, |
| [10] |
孙珊珊,满都呼,伟军, |
| [11] |
董克池,郝媛媛,董瑞, |
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
董光,狄威,程武学 . 基于低空遥感的若尔盖草地鼢鼠鼠害信息提取方法及对比研究[J]. 四川师范大学学报(自然科学版),2022,45(1):110-118. |
| [17] |
张涛,杜健民,张海军, |
| [18] |
马涛,郑江华,温阿敏, |
| [19] |
孙迪,倪亦非,陈吉军, |
| [20] |
杜波波,冀超,王大伟, |
| [21] |
堂格斯,包玉龙,孙长青, |
| [22] |
汪超,张韬 . 基于MODIS遥感影像辉腾锡勒草原NDVI变化与气象因子相关性研究[J]. 内蒙古农业大学学报(自然科学版),2021,42(5):32-45. |
| [23] |
|
| [24] |
皮伟强,杜健民,陈程, |
| [25] |
张学工 . 关于统计学习理论与支持向量机[J]. 自动化学报,2000,26(1):32-42. |
| [26] |
李蓉,叶世伟,史忠植 . SVM-KNN分类器:一种提高SVM分类精度的新方法[J]. 电子学报,2002,30(5):745-748. |
| [27] |
张昊,潘新,罗小玲, |
| [28] |
刘洋洋 . 基于植被指数的无人机多光谱影像鼢鼠土丘提取评价[D]. 呼和浩特:内蒙古农业大学,2023. |
| [29] |
李鹤 . 基于无人机高光谱遥感的典型荒漠草原的鼠洞识别研究[D]. 呼和浩特:内蒙古农业大学,2021. |
| [30] |
朱相兵,毕玉革,刘浩, |
| [31] |
花蕊,董瑞,包达尔罕, |
| [32] |
包宝小,潘新,马玉宝, |
| [33] |
张兵 . 高光谱图像处理与信息提取前沿[J]. 遥感学报,2016,20(5):1062-1090. |
| [34] |
周晓琳,安如,陈跃红, |
| [35] |
林忆馨 . 无人机多光谱遥感在两种鼢鼠危害调查中的应用研究[D]. 呼和浩特:内蒙古农业大学,2022. |
内蒙古自治区自然科学基金项目(2023MS03025)
国家自然科学基金项目(32060256)
国家自然科学基金项目(32060395)
内蒙古自治区科技计划项目(2021GG0108)
内蒙古自治区科技重大专项项目(2021ZD0006)
/
| 〈 |
|
〉 |