像素级与对象级结合的变化检测方法

Combination of Pixel-based and Object-based Change Detection Method

提出一种像素级与对象级结合的变化检测方法。通过像素级的主成分分析法获得代表差异信息的主成分影像;再用多尺度分割算法分割主成分影像,得到最终的变化检测结果。最后利用landsat影像进行实验,不仅可以有效抑制由误差造成的伪变化,同时提取的变化区域更加完整。

In view of the drawbacks of the pixel-based and object-based change detection methods, we proposed a hybrid approach for change detection in this paper. Firstly, the principal component image was found by pixel-based PCA. And then, the principal component image was segmented by multi-solution segmentation method, and got the final result. The experimental results show that the proposed method not only can effectively restrain the pseudo changes caused by errors, but the proposed final change area is more complete.