面向无人机影像的改进FNEA分割方法  

Improved FNEA Segmentation Method for UAV Images

作　　者：张瑞峰[1] 闫超德[1,2] 罗先学[3] 袁观杰 叶勇超 ZHANG Ruifeng;YAN Chaode;LUO Xianxue;YUAN Guanjie;YE Yongchao(School of Water Conservancy Engineering,Zhengzhou University,Zhengzhou 450001,China;Yellow River Institute for Ecological Protection&Regional Coordination Development,Zhengzhou University,Zhengzhou 450001,China;Zhengzhou Urban Planning Design&Survey Research Institute,Zhengzhou 450001,China)

机构地区：[1]郑州大学水利科学与工程学院,河南郑州450001 [2]郑州大学黄河生态保护与区域协调发展研究院,河南郑州450001 [3]郑州市规划勘测设计研究院,河南郑州450001

出　　处：《地理信息世界》2022年第3期1-6,共6页Geomatics World

基　　金：国家自然科学基金(41671455);中国工程院重大咨询研究项目(2021-149-1)。

摘　　要：针对无人机影像光谱信息量不足导致的影像分割精度较低和同类地块过度分割问题,本文设计了一种面向无人机影像的改进FNEA分割方法。首先,利用特征提取方法,构建原始RGB影像的不同纹理和植被指数特征影像;然后,基于改进分离阈值法选择最佳植被指数和纹理特征;最后,将最佳特征与原始RGB数据融合,采用FNEA算法进行影像分割。将改进FNEA方法与多种分割方法对比,结果表明改进FNEA方法的分割精度更高,同一地块内过度分割可得到较好控制,适合无人机影像的分割。To solve the issues of low image segmentation accuracy and over-segmentation of same blocks caused by lack of spectral information of UAV image,this paper proposes an improved FNEA segmentation method for UAV images.Firstly,different texture and vegetation index feature images of the original RGB images are constructed by feature extraction.Secondly,the best vegetation index and texture feature in UAV images are selected by an improved separation threshold method.Finally,the FNEA algorithm is used for image segmentation after fusing the best features with the original RGB data.By comparing the segmentation experiments of the improved FNEA method with multiple segmentation methods,the experimental results show that,the proposed method not only has higher segmentation accuracy,but also well solves the problem of over-segmentation in the same blocks,which means FNEA is more suitable for the segmentation of UAV images.

关 键 词：无人机影像 特征选择 改进分离阈值法 FNEA影像分割算法 影像分割 

分 类 号：P2[天文地球—测绘科学与技术]