针对高空间分辨率卫星遥感图像上道路绿地的特点,提出一种基于道路绿地形状和空间分布特征识别道路边线和中心线的方法。在基于NDVI提取绿地信息的基础上,根据道路绿地的形状特征(细长),采用面积、主轴方向、紧致度、矩形度和长宽比等形状指数,区分道路绿地与其他绿地;根据道路绿地之间以及道路绿地和道路之间的空间关系特征,同时通过生成已知道路绿地区域的缓冲区,分隔那些与非道路绿地连成一片的道路绿地;最后,根据道路绿地的方向和距离特征提取出道路边线和中心线。
In consideration of characteristics of mad greenbelts in high spatial resolution satellite imagery, this paper proposes a method combining features of shape with those of spatial distribution of road greenbelts to recognize road edges and road centerlines. The method first extracts greenbelt information from a NDVI image, then differentiates road greenbelts from other greenbelts by means of linear features of road greenbelts and their related shape indices such as area, principal axis direction, compactness, rectangularity, and ratio of width and length. After that, the road greenbelts adjacent to other greertbelts are separated by generating buffers of known mad greenbehs as well as according to the spatial relationships between different greenbelts and between mad greenbelts and other greenbehs. Finally road edges and road centerlines are extracted in accordance with the direction and distance characteristics of road greenbehs.