面向对象变化向量分析的遥感影像变化检测

摘要

In order to solve the problem of low change detection precision for high spatial resolution remote sensing imagery using pixel-based change vector analysis method,an object-oriented change vector analysis(OCVA)method was proposed.Image segmentation was used to get image objects according to the images at two times.The features of the objects were extracted by the histogram with the objects.Change vectors of the objects were built by the histogram distance measure using histogram intersection method.The mold of the change vector was calculated by the weighted combination of the elements of it.The changed/unchanged labels of the objects were decided by the change detection threshold obtained by maximum entropy principle.The experimental results on QuickBird and Ikonos images showed that:compared with the pixel based change vector analysis,the proposed method can get higher change detection accuracy for high spatial resolution imagery,and the correct rate reaches 0.92 and 0.90 respectively.%为了解决基于像元的变化向量分析法在高分辨率遥感影像变化检测中精度低的问题,提出了一种面向对象变化向量分析的遥感影像变化检测方法.综合2个时期的遥感影像,首先通过影像分割获取像斑,其次提取直方图作为像斑的特征向量,再次采用直方图相交法度量2个时期像斑直方图之间的距离,构建像斑的变化向量,然后利用加权组合的方法计算像斑变化向量的模,最后依据最大熵原理获取变化检测阈值,对像斑进行变化/未变化判别.在QuickBird及Ikonos遥感影像上的实验表明:在高分辨率遥感影像变化检测中,与基于像元的变化向量分析法相比,该方法变化检测的精度较优,变化检测的正确率分别达到了0.92与0.90.