Unsupervised Image Change Detection Based on 2-D Fuzzy Entropy

Change detection in images of a given scene acquired at different times is one of the most interesting topics of image processing. A new change detection method based on 2-D fuzzy entropies is proposed in this paper to detect change area of the difference image. First, the best segmentation direction of 2-D histogram formed by pixel gray levels and the local average gray levels is found by using Fisher criterion. Then, a kind of new 2-D membership function is defined based on the best segmentation direction, which is used to obtain the optimal membership function by searching 2-D maximal fuzzy entropy. Finally, the image change area is detected by using the optimal membership function. The theoretical analysis and experiment results show that the proposed method has predominant change detection performance.