摘要
通过对高空间分辨率遥感图像(简称高分图像)中的居民区纹理结构信息的分析,提出了一种基于Gabor滤波和局部特征点密度的居民区提取方法。该方法首先对高分图像进行多方向Gabor滤波,得到多个方向的幅值信息,并通过阈值处理和筛选后处理获取图像的特征点;然后对特征点求取局部密度,获取居民区的范围;再用数学形态学变换进行细微处理,最终提取出图像中的居民区。以World View2真彩色图像为实验数据对不同方法进行验证及对比分析的结果表明,该方法具有较高的提取精度和计算效率。
To tackle the problem of urban area detection using high - resolution remote sensing images, this paper proposes a method based on Gabor filtering and density of local feature points by analyzing the residential area texture of high resolution image. For obtaining the amplitude information in multiple directions, the Gabor filtering was used firstly, and then the image feature points were extracted by subsequent processing of amplitude images. By computing the density of local feature points, the initial residential areas could be obtained. With further mathematical morphology transformation of the areas, the results were optimized ultimately. In the experiments, two WorldView2 data were used to validate the different methods. A comparative analysis with other methods shows that the method proposed in this paper has higher extraction accuracy and computational efficiency for urban area detection.
出处
《国土资源遥感》
CSCD
北大核心
2015年第3期59-64,共6页
Remote Sensing for Land & Resources
基金
国家重点基础研究发展计划"973"项目"高分辨率遥感影像的目标特征描述与数学建模"(编号:2012CB719903)
高分辨率遥感交通应用示范项目"高分综合交通遥感应用示范系统先期攻关"(编号:07-Y30A05-9001-12/13)
四川省测绘地理信息局科技计划项目"基于规则驱动的城市三维快速建模技术研究"(编号:J2014ZC02)共同资助
关键词
高空间分辨率遥感图像
多方向
GABOR滤波
局部特征点密度
high resolution remote sensing image
multi - oriented
Gabor filtering
density of local feature points
