Local Saliency Extraction for Fusion of Visible and Infrared Images

In this paper, a local saliency extraction-based dual-band image fusion algorithm is proposed. Combing the variable computational windows, the local gray distance is designed for saliency analysis. And saliency map is further obtained by considering spatial weight. For dual-band image fusion, firstly, we design several local windows named different levels, and get the corresponding saliency maps. Secondly, achieve weighted fusion under different levels with saliency maps. Finally, all fused images are compounded into one fused result. According to experimental results, the proposed method could produce a fused image with good visual effect, preserving even enhancing the details effectively. Comparing with other seven methods, both subjective evaluation and objective metric indicate that the proposed algorithm performs best.