利用多特征融合的面向对象的高分辨率遥感影像变化检测

Object-oriented change detection for high resolution remote sensing images by integrating multiple features

针对高分辨率遥感影像变化检测中的一些难点和传统基于像元的变化检测方法的局限性,提出了一种多特征融合的面向对象变化的检测方法。为保证图斑的空间位置对应,首先利用改进的图像分割算法对两幅影像进行联合分割,然后统计图斑的多种特征,依据该特征进行变化向量分析,得到图斑的变化和非变化类型。最后,利用Quick Bird影像验证了该算法的可行性,检测结果明显优于传统基于像素的检测方法。

In order to solve several difficulties of change detection for high-resolution remote sensing images and the limitation of traditional pixel-based change detection method,the paper presents a novel objectlevel change detection method by integrating multiple features. First of all,in order to ensure the spatial location of every polygon matches precisely,a modified graph-based segmentation algorithm was adopted to segment two image,then calculating the specified features of the object,and according to the features,the change vector analysis method was used to obtain the change or unchanged types of the corresponding polygons. Finally,to verify the effectiveness of the algorithm,experiments were carried out on Quick Bird image,and the result shows that the object-based method performs very well and is obviously superior to the pixel-level change detection methods.