A dual-weighted polarization image fusion method based on quality assessment and attention mechanisms

This paper proposes a dual-weighted polarization image fusion method based on quality assessment and attention mechanisms to fuse the intensity (S0) degree of linear (DoLP). S0 has high contrast clear details, DoLP an outstanding ability characterize properties, so can achieve effective complementation superior information. We decompose into base layers detail layers. In layers, we build unit combining information entropy, no-reference assessment, local energy ensure fused natural visual perception; in layer, first extract depth features using pre-trained VGG19, then construct enhancement space channels, finally effectively improve preservation edge contours image. The proposed is able perceive retain sufficiently obtain desirable results. Comparing nine typical methods two publicly available own datasets, experimental results show that outperforms other comparative algorithms both qualitative comparison quantitative analysis.