Change Detection in Remotely Sensed Images Based on Image Fusion and Fuzzy Clustering

摘要

Change detection in remotely sensed images involves a multi-temporal dataset, an algorithm to detect the change and classification of the difference image into changed and unchanged regions. The quality of results mainly depends upon the algorithm used for change detection. In this paper, a Modified Discrete Wavelet Transform (DWT) based image fusion method for change detection has been proposed. The changed and unchanged areas are segmented by fuzzy c means clustering (FCM). The algorithm has been implemented in MATLAB R2013 on two datasets. The first dataset belongs to the Bi-temporal images of the area of Reno Lake. The first image was captured on August 5, 1986 and the second image was captured on August 5, 1992. The second dataset used is from the city of Ottawa. The results are compared based upon various parameters like Percentage correct classification or Accuracy (PCC) and Kappa coefficient (K_c). The qualitative and quantitative results show that the accuracy and Kappa value of proposed method is higher than the pixel averaging DWT based fusion method.