Enhancing Infrared Images Via Multi-Resolution Contrast Stretching and Adaptive Multi-Scale Detail Boosting
Haoxiang Lu,Zhenbing Liu,Xipeng Pan,Rushi Lan,Wenhao Wang
DOI: https://doi.org/10.1007/s00371-022-02765-y
2024-01-01
Abstract:Infrared imaging technology has attracted numerous interests in military, security, transportation, etc. However, infrared images suffer from low contrast and blurry detail, limiting its application. To handle these issues, this paper presents a novel approach based on multi-resolution contrast stretching and adaptive multi-scale detail boosting. Specifically, the hybrid-spatial Gamma function can create the high-brightness feature map; multi-resolution analysis fully discovers the information in different screen resolutions for reconstructing the high-contrast feature map, i.e., the weighted adaptive limited contrast histogram equalization stretches the image’s approximate map and the hybrid Gamma function and histogram processes the detailed maps at different directions; the edge-preserving Gaussian blurring function can fully explore the details at different scale space for yielding the feature map with clearer details. Finally, the Laplace pyramid fuses different feature maps obeying the image-matching degree and average gradient to reconstruct pleasing visual images. Extensive experiments have shown that our method has an average increased by 13.0411, 1.6737, 23.3651, and 20.8085 in terms of average gradient (AG), information entropy (IE), enhancement by IE (EME), and image sharpness (IS), respectively, which is a proven approach for infrared image enhancement.