遥感图像城市道路细节特征提取及增强方法研究  

Study of Urban Roads Detail Extraction and Enhancement for Remote Sensing Image

作　　者：孙海燕 

机构地区：[1]北方工业大学计算机学院,北京100144

出　　处：《工业技术创新》2018年第1期87-90,共4页Industrial Technology Innovation

摘　　要：高分辨率遥感图像能够呈现出更加丰富的细节特征,但在图像处理过程中,这些细节特征通常被当作噪声,从而使得图像信息的利用率降低。采用数学形态学方法提取城市道路遥感图像中较亮的细节特征,采用Frangi滤波器增强开运算结果中的线性元素,使用张量投票方法增强诸如车辆等细节特征间的方向关系。对车辆密集的情况进行实验,道路主方向突出效果较好,表明提出的算法有利于复杂场景下城市道路细节特征的提取及增强。High resolution remote sensing images contain a lot of details, but they are usually treated as noise during image processing, thus the information from images cannot be taken full use of. The morphology method is adopted to extract the bright details from the urban road remote sensing image, and then, the Frangi filter and tenser voting methods are executed to enhance the linear objects in the open operation results and stand out the principle direction of details like cars. Experiment is carried out under the condition of heavy traffic, showing that the principle direction of the road is stood out, indicating that the proposed algorithm can benefit the urban roads extraction and enhancement under complex scenes.

关 键 词：数学形态学 Frangi滤波器 张量投票 遥感图像 特征提取 

分 类 号：TN431.1[电子电信—微电子学与固体电子学]