基于无人机遥感与面向对象法的田间渠系分布信息提取

发布时间：2018-06-07 09:05

本文选题：灌区渠系分布 + 信息提取　；参考：《农业机械学报》2017年03期

【摘要】：针对目前农田灌排系统识别研究中遥感影像分辨率不足,难以提取田间毛渠且对无水或少水灌排沟渠识别不足等问题,以内蒙古河套灌区磴口县坝塄村为研究区域,利用固定翼无人机搭载520~920 nm多光谱相机进行航拍试验,采用基于面向对象法的特征组合分层分类的提取方法对获取的高分辨率单幅多光谱影像数据进行解译,采用分割阈值为65、合并阈值为90的遥感影像最佳分割参数。利用含水田间毛渠和无水、少水田间毛渠在光谱、几何、空间关系等特征参量中表现出的与其它地物的特异性,建立不同分类层次的规则提取田间毛渠分布信息。提取结果表明,由于水体对近红外波段光谱的强烈吸收,含水毛渠提取效果很好,精度达到97.8%;无水、少水田间毛渠提取精度为75.7%。无人机遥感技术和面向对象法的特征组合分层分类方法为灌区田间渠系识别提供了一种新途径。
[Abstract]:In view of the problems of insufficient resolution of remote sensing image in the research of farmland irrigation and drainage system recognition, it is difficult to extract the field furrow and to identify the drainage ditch without water or water, etc., taking Dengkou County, Dengkou County, Inner Mongolia Hetao Irrigation District as the research area, as the research area. In this paper, the fixed-wing UAV (UAV) is used to carry out aerial photography test with 520m 920nm multispectral camera. The extraction method of feature combination and hierarchical classification based on Object-Oriented method is used to interpret the obtained high-resolution single multi-spectral image data. The optimal segmentation parameters of remote sensing images with a threshold of 65 and a combined threshold of 90 are adopted. Based on the characteristic parameters of spectral, geometric and spatial relationship, the distribution information of the field wool canal was extracted by the rules of different classification levels, such as spectral, geometric, spatial relationship and so on, using the characteristic parameters of water-bearing field canal and water-free field canal and lack of water as well as the characteristic parameters such as spectral, geometric and spatial relations. The results showed that the extraction accuracy of the aqueduct was 97.8because of the strong absorption of the near infrared spectrum in the water body, and the precision of extracting the furry canal with little water was 75.7in the field without water. UAV remote sensing technology and Object-Oriented method provide a new approach for field canal system identification in irrigation area.
【作者单位】：西北农林科技大学机械与电子工程学院;西北农林科技大学水土保持研究所;
【基金】：科技部国际合作项目(2014DFG72150) 杨凌示范区工业项目(2015GY-03)
【分类号】：S274;S127

【相似文献】