Detection of changes from Satellite Images Using Fused Differene Images and Hybrid Kohonen Fuzzy C-Means Sigma

An unsupervised change detection method based upon wavelet fusion and Hybrid Kohonen FCM-σ is presented in this paper. Principle component analysis of the bi-temporal satellite images of same scene is computed for dimensionality reduction. The log ratio & absolute difference images of PC1 are fused together using wavelet transform. Further HKFCM-σ classifier is used to classify the satellite image. Since the focus is on change detection thus the images are classified into two classes only representing change and unchanged areas. The simulation results of proposed method show that it can be effectively used for change detection. The result of the proposed technique is furthermore equalled with the existing change detection techniques available in the literature and it is observed that the proposed technique outperforms the other methods.