基于多因子几何匹配的AI提取路网属性信息重建方法

利用AI(artificial intelligence)技术可从遥感影像上快速提取矢量数据,尤其可以获取实时性较好的矢量路网数据,但提取的数据没有属性信息;而已有的众源数据如OSM(open street map)路网具有开源、属性信息丰富等特点,但现势性相对于提取路网较低.针对上述情况,以AI提取路网为基准数据,OSM路网为匹配数据,将一种基于多因子几何匹配算法用于路网匹配中,并在匹配后引入匹配度的概念,以最优匹配对象进行属性重建.实验结果表明能有效地对AI提取路网的属性信息进行重建,并基于此开发了一套路网属性信息重建系统,在国家全球测图项目中投入使用.

Attribute information reconstruction method of road network extracted by AI based on multi-factor geometric matching

Using AI technology can extract vector data from remote sensing images rapidly, especially the vector road network data with good real-time performance, but the extractive data has no attribute information. Existing multi-source data such as OSM road network data are open source and rich in attribute information, but the present situation is lower than that of road network extraction. According to the above situation, the AI extractive road network as benchmark data, the OSM network as the matching data, this paper applied a multi-factor geometric matching algorithm to the network matching process, and introduced the concept of matching degree after the match, carried out attribute reconstruction with the best matching object. The experimental results show that it can effectively reconstruct the AI extractive road network attribute information, and based on this, it developed a road network attribute information reconstruction system, which applied in the national global mapping project.