Information-theoretic Heterogeneity Analysis for Evaluating the Performances of SAR Image Fusion Techniques

An improved evaluation method based on information theory for the synthetic aperture radar (SAR) images fusion performance is investigated in this paper. Through employing relative entropy, weighted entropy and other information-based variables to present the information transferring in the fusion, it can assess the performance of different fusion techniques effectively. However, the inevitable effect of the speckle noise on the SAR images will make the information calculation encounter inaccuracy. Therefore, an approach to extract the information which can reflect the spatial dependencies in the observed SAR data information is needed. This paper applied an information-theoretic heterogeneity feature measurement, which can efficiently emphasize the spatial dependencies. Experimental results carried out on different SAR images and the fusion images demonstrate the applicability of the proposed idea in SAR images fusion performance evaluation