基于机载LiDAR数据和遥感影像的复杂建筑物提取  

Complex Building Extraction Based on Airborne LiDAR Data and Remote Sensing Image

作　　者：党涛 党丹凤 周立勇 任建平 

机构地区：[1]兰州大学资源环境学院,甘肃兰州730000 [2]西安测绘总站,陕西西安710054 [3]湖北省宜昌地质勘探大队,湖北宜昌443100

出　　处：《地理空间信息》2018年第9期69-71,74,共4页Geospatial Information

基　　金：国家自然科学基金资助项目(41271360)

摘　　要：综合利用机载LiDAR数据和高分辨率遥感影像数据优势,提出了一种面向对象的分层分类提取复杂建筑物的新方法。首先根据坡度强度信息,将影像分割成高、中、低坡度目标;再采用阈值法进行陡峭区、地表面和建筑物的初始分类;最后根据邻近对象光谱相似性原则,对陡峭区进行多尺度分割,并结合光谱、形状和空间关系等特征,基于模糊分类对建筑物提取结果进行优化。实验表明,该方法提取的建筑物信息精度较高,轮廓边缘相对完整。Based on the advantages of airborne LiDAR data and high resolution remote sensing image data, we present a new extraction method for complex building. Firstly, based on the slope intensity information, we separated the DSM_slope image into high, medium and low slope objects. And then, we used the threshold method to classify the steep, ground and building objects initially. Finally, according to the similar principle of adjacent object, we used multi-resolution segmentation to reshape the steep objects, and based on the characteristics of the spectrum, shape and spatial relationship of the high resolution images, optimized the building extraction results based on the fuzzy classification. The experimental results show that the proposed method is of high precision, and the contour edge is relatively complete.

关 键 词：机载LIDAR 高分辨率影像 建筑物提取 面向对象 

分 类 号：P237[天文地球—摄影测量与遥感]