Multi-scale infrared image enhancement based on non-uniform weighted guided filtering

Enhancement methods have become indispensable due to low-contrast and blurred measurements of infrared imaging systems. However, most existing infrared image enhancement methods suffer from less balance between the high-frequency features and robustness to noise. Here, a multi-scale infrared image enhancement algorithm based on non-uniform weighted guided filtering (NWGIF) is proposed to enrich details as well as reduce noise. Our designed framework utilizes NWGIF for multi-scale image decomposition to separate features in the single base layer and multi-scale detail layers. Then, an adaptive brightness correction model integrated with the defogging algorithm adjusts the brightness of the base layer. In addition, the high-frequency features hidden in multi- scale detail layers are enhanced with the help of a differential gain function based on the directional gradient operator. Thanks to the weighted fusion of the single base layer and multi-scale detail layers, our method achieves a high-quality enhancement with an average natural image quality evaluator (NIQE) of 4.48. We experimentally demonstrate that our method realizes a higher-fidelity detail enhancement with better robustness to Gaussian noise than the six existing classical methods. The high-quality results could provide potential application support in special imaging tasks, such as target recognition and tracking.