Change detection from remotely sensed images based on a decision theoretic method

In this paper, a change detection method for remotely sensed satellite images based on a decision theoretic method is proposed. The proposed method works in two stages. In the first stage, a difference image was computed using change vector analysis (CVA) technique. For multispectral images, CVA technique works well as it uses all the bands of two multidate satellite images to compute the difference image. In the next stage, the algorithm aims to classify the difference image into two classes: changed and unchanged, in such a way as to minimize the probability of misclassifications. Finally, a decision function based on the mean and covariance matrix of the changed and unchanged classes is used to build a change detection map showing changed and unchanged areas. The proposed algorithm was compared quantitatively and qualitatively with other existing state of the art methods. The results showed that the proposed method outperforms the other methods.