基于最优尺度模型的建筑物信息提取

Building Information Extraction Based on Optimal-scale Model

建筑物信息在数字城市建设过程中具有十分重要的意义和作用,从高分辨率遥感影像中提取建筑物信息已成为研究热点之一。本文结合当前现有的理论与方法,提出最优尺度的面向对象的方法对建筑物信息进行提取。首先采用多尺度分割算法对影像进行分割,然后采用最优尺度计算模型选取最优分割尺度,在分割的基础上构建建筑物提取知识库,最后采用最近邻方法将影像中的建筑物提取出来。实验结果表明,本文提出的方法能快速有效地提取出建筑物信息,且精度可靠。

The building information plays a very important role in the construction of the digital city.The extracting method of the building information from high-resolution remote sensing images has been the hot research topic.With the current existing theories and methods,this paper proposes an object-oriented building information extraction method based on optimal-scale model.Firstly,it used multi-scale segmentation algorithm to segment image,and chose the optimal segmentation scale by optimal scale computing model.Then it constructed the knowledge base for extracting the building information with the image segmentation.Finally,it extracted the building information from the images by the nearest neighbor method.The result shows that it can extract building information by this method quickly and efficiently,and the accuracy is reliable.