An Unsupervised Change Detection Based on Clustering Combined with Multiscale and Region Growing

In this paper, a novel approach is proposed for unsupervised change detection of multitemporal remote sensing images. The proposed method is able to produce the change detection result on the difference image without a priori assumptions .Firstly, the difference image which is acquired from multitemporal images. Mean shift algorithm is used to reduce noise of difference image and fake change. Then stationary wavelet transform (SWT) is used to extract feature vector of each pixel .The final change detection map is achieved by k-means clustering combined with region growing. The comparisons with the state-of-the-art change detection methods are provided.