摘要
为提高园林景观设计中高分辨率遥感影像道路提取的精度及效果,提出一种融合SVM的高分辨率遥感影像道路提取方法。该方法首先结合Mean Shift算法与数学形态学运算(简称MS-MMO)进行影像阴影提取;再根据阴影提取结果对原始影像阴影区域进行亮度补偿后输入SVM,得到初步提取的道路图像;然后利用高斯滤波算法进行图像平滑处理,利用边缘滤波、纹理滤波等算法去除图像中的非道路区域,得到道路区域提取图;最后基于张量投票提取道路中心线,基于“交点”搜索方法去除道路中心线上的毛刺,完成道路提取。实验结果表明,MS-MMO的具有较好的阴影提取精度及效果;根据MS-MMO输出的阴影提取结果对原始影像阴影区域进行亮度补偿后,道路提取的整体性能更高;融合SVM的高分辨率遥感影像道路提取方法提取的道路完整性、正确性、质量分别达到92.4%、92.7%、89.0%,道路提取性能较好,且道路具有连通属性,在该方法提取的道路图像上进行园林景观设计,可有效提升道路植物配置效果。
To improve the accuracy and effectiveness of road extraction from high-resolution remote sensing images in landscape design,a road extraction method based on SVM fusion is proposed.Firstly,the method combines Mean Shift algorithm and mathematical morphology operation(MS-MMO for short)to extract shadow from images;Then,based on the shadow extraction results,the brightness of the shadow area in the original image is compensated and input into SVM to obtain the preliminary extracted road image;Then,the Gaussian filter algorithm is used to smooth the image,and the edge filter,texture filtering and other algorithms are used to remove the non road areas in the image to obtain the road region extraction image;Finally,the road centerline is extracted based on tensor voting,and the burrs on the road centerline are removed using the“intersection”search method to complete the road extraction.The experimental results show that MS-MMO has good shadow extraction accuracy and effectiveness;After compensating the brightness of the shadow area in the original image based on the shadow extraction results output by MS-MMO,the overall performance of road extraction is higher;The road integrity,accuracy,and quality extracted by the high-resolution remote sensing image road extraction method fused with SVM reached 92.4%,92.7%,and 89.0%,respectively.The road extraction performance is good,and the roads have connectivity properties.Landscape design on the road images extracted by this method can effectively improve the effectiveness of road plant configuration.
作者
李华
LI Hua(Xi’an International University,Xi’an 710077,China)
出处
《自动化与仪器仪表》
2024年第2期15-19,共5页
Automation & Instrumentation
基金
西安美术学院校级人文社会科学研究项目《二十世纪中国“红色美育”发展历程研究》(2021XK025)。
关键词
均值漂移算法
数学形态学运算
支持向量机
道路提取
滤波算法
mean shift algorithm
mathematical morphology operation
support vector machine
road extraction
filtering algorithm
