基于高分三号卫星SAR影像的城市建筑区提取

Urban Building Area Extraction Based on GF-3 Satellite SAR Images

城市研究对于城市的合理规划与发展有着重要的意义,建筑区提取是城市研究必不可少的工作。合成孔径雷达(SAR)具有全天时、全天候、高分辨率等特点,已成为城市研究的重要数据源之一,我国最新发射的高分三号合成孔径雷达卫星尤其引人关注。在分析建筑区散射机制和SAR影像特征的基础上,本文针对高分三号SAR影像,筛选出最优纹理特征组合,确定了合适的窗口尺寸,并以盐城地区高分三号SAR影像为例,综合利用灰度和纹理特征进行了建筑区提取。研究表明:综合利用具有较大巴氏距离值的3个纹理特征和灰度特征同样可以在高分三号影像上实现建筑区的自动提取,提取结果的精度接近72%,比单纯使用灰度信息的提取精度有所提高,为高分三号卫星数据在城市研究中的应用提供了参考。

Urban research is of great significance to the rational planning and development of cities.The extraction of building areas is an indispensable task in urban research.Synthetic aperture radar(SAR)has allweather,high resolution and other characteristics.It has become one of the important data sources for urban research.China latest launched of GF-3 synthetic aperture radar satellite is especially attractive.Based on the analysis of the scattering mechanism and SAR image features of the building area,this paper selects the optimal combination of texture features for the GF-3 SAR images and determines the appropriate window size.Taking the GF-3 SAR image of Yancheng area as an example,the construction area is extracted by using comprehensive utilization of grayscale and texture feature.The result shows that the building area can be automatically extracted through comprehensive utilization of three texture features and gray features with large Bhattacharyya distance value on GF-3 images.The accuracy of the extraction results is close to 72%,and the results provides a reference for the application of GF-3 satellite data in urban research.