Adaptive low light visual enhancement and high-significant target detection for infrared and visible image fusion

Wenxia Yin, Kangjian He, Dan Xu, Yingying Yue, Yueying Luo

Abstract: Infrared and visible image fusion aim to obtain a fused image with salient targets and preserve abundant texture detail information as much as possible, which can potentially improve the reliability of some target detection and tracking tasks. However, some visible images taken from low-illumination conditions are subjected to losing many details and cannot obtain a good fusion result. To address this issue, we proposed a novel adaptive visual enhancement and high-significant targets detection-based fusion scheme in this paper. First, a bright-pass bilateral filter and adaptive-gamma correction-based algorithm are proposed to enhance the visible image adaptively. Second, an iterative guided and infrared patch-tensor-based algorithm are proposed to extract the infrared target. Third, an efficient hybrid \(\ell_{1} - \ell_{0}\) model decomposes the infrared and visible image into base and detail layers and then fuses them by weight map strategy. The final fused image is obtained by merging the fused base layers, detail layers, and infrared targets. Qualitative and quantitative experimental results demonstrate that the proposed method is superior to 9 state-of-the-art image fusion methods as more valuable texture details and significant infrared targets are preserved. Supplemental material and codes of this work are publicly available at: https://github.com/VCMHE/BI-Fusion.