Saliency-based geometry measurement for image fusion performance

A novel quality metric for image fusion using saliency-based geometry measurement is presented. The method provides a quantification of salient structural information transferred from each source into the fused image. Viewing the multiple measures as a point for an image fusion technique, the final fusion performance is evaluated via computing how far the distance is from the result point to an ideal point with perfect value in each source quantification dimension. Experimental results indicate that the proposed metric outperforms other popular performance assessment methods for image fusion.