摘要
针对基于像素的道路提取方法的不足,使用一种基于超像素分割算法(Simple Linear Iterative Clustering,SLIC)和自适应阈值分割算法(OTSU算法是由日本学者OTSU于1979年提出的一种对图像进行二值化的高效算法)相结合的道路提取方法,可以较好地解决在遥感图像中分辨率较高所造成的非道路地物对目标的噪声影响。该方法使用SLIC超像素分割算法对影像进行分割处理,再用改进的K-means聚类算法对分割后的超像素影像进行非监督分类,根据GVI值对分类后的影像中的植被及水体信息进行过滤,对过滤后的影像进行基于OTSU的分割,最后对分割影像进行后处理获得完整道路网。经过定性和定量分析后得出,此方法在道路提取上有较好的表现。
In view of the shortcomings of pixel-based road extraction methods,this paper used a road extraction method based on SLIC segmentation and OTSU adaptive threshold,which could better solve the impact of non-road surface objects on the target noise caused by high resolution in remote sensing images.The method used SLIC super pixels segmentation algorithm for image segmentation,then with the modified K-means clustering algorithm to unsupervised segmentation after super pixel image classification,according to the values of GVI to vegetation and water body information of image classification after filtering,the filtered image segmentation based on OTSU algorithm,and finally affected segmentation after post-processing for the entire road network.Through qualitative and quantitative analysis,it is concluded that this method had a good performance in road extraction.
作者
吴激涛
刘荣
WU Jitao;LIU Rong(Faculty of Geomatics,East China University of Technology,Nanchang Jiangxi 330013,China)
出处
《北京测绘》
2021年第5期590-594,共5页
Beijing Surveying and Mapping
基金
国家自然科学基金(42004072)。
