一种基于影像分割的城市建筑物变化检测方法

A Method for Urban Building Change Detection Based on Image Segmentation

针对高分辨率遥感影像建筑物变化检测精度不高的问题,本文提出了一种改进城市建筑物变化检测方法。首先通过提取像元顶点构造像元图集,并以长宽比与矩形度作为变化检测测度,对后一期影像进行影像分割,识别建筑物轮廓对象。将建筑物轮廓进行几何关系筛选,完成建筑物变化信息提取。实验表明,该方法具有较高的变化检测精度,可明显削弱光照条件和成像角度对建筑物变化检测精度的影响,是一种普适性较强的城市建筑物变化检测方法。

Aiming at the low accuracy of building change detection in high resolution remote sensing images,this paper proposed an improved urban building change detection method.Firstly,the pixel atlas was constructed by extracting pixel vertices,and the length width ratio and rectangle ratio were used as the change detection measure,and the later image was segmented to identify the outline of the building.The geometric relationship of buildings was screened and the information of changed building was extracted.The experiment showed that the method has a high accuracy of change detection,and can obviously weaken the influence of illumination conditions and imaging angles on the detection accuracy of the building changes,it is a universal method of urban building change detection.