文章摘要

文章信息/Info

Title:: Comparison of Algorithms for Surface Buildings Extraction Based on UAV Images:Taking Qiongharajun Village in Atushi City,Xinjiang as an Example

Author(s):: WU Haolin; NIE Gaozhong; FAN Xiwei; WEI Benyong; AN Jiwen; (Institute of Geology,China Earthquake Administration,Beijing 100029,China)

Keywords:: pre-assessment of earthquake disaster losses; UAV; surface buildings extraction; morphological imageprocessing

摘要:: 选取新疆阿图什市下辖的琼哈拉峻村为研究区,将小型旋翼无人机拍摄的图像作为数据源,分别采用面向对象以及面向像元2种影像分析方法对研究区的房屋建筑进行提取,并对2种算法的提取结果进行对比,分析了各自的优势。结果表明:面向对象方法可以有效地去除椒盐噪声对分类带来的影响,保证房屋形态的完整性,但影响内部相似的光谱、纹理信息若对应多种物体则会导致影像对象的错分。在面向像元的提取方法中加入了改进的数学形态学算法,可有效的抑制椒盐噪声,保持建筑物边缘的连续性与完整性,较好地解决了面向对象方法中部分农田与房屋出现错分的问题。

Abstract:: Taking Qiongharajun Village in Atushi City,Xinjiang as the research area,using the image taken by the small-scale rotorcraft UAV as the data source,we extract the housing buildings in the research area by object-oriented and pixel-oriented image analysis methods,compared the results of the two algorithms,and analyzed their respective advantages. The results show that the object-oriented method can effectively eliminate the impact of salt and pepper noise on classification and ensure the integrity of the housing morphology. However, if the similar spectral and texture information in the image corresponds to a variety of ground objects, the image object will be misclassified. The improved mathematical morphology algorithm is added to the pixel-oriented extraction method, which can effectively suppress salt and pepper noise, maintain the continuity and integrity of the building edge, and solve the problem of some farmland and houses in the object-oriented method.