A System for Fusing Color and Near-Infrared Images in Radiance Domain

Kim C Ng, Jinglin Shen, Chiu Man Ho; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, 2021, pp. 2021-2030

Abstract


We designed and demonstrated a system that fused color and near-infrared (NIR) images in the radiance domain. The system is designed to enhance image quality captured in outdoor environments, especially in hazy weather conditions. Previous dehazing methods based on RGB-NIR fusion exist but have rarely addressed the issue of color fidelity and potential see-through effect of fusing with NIR image. The proposed system can dehaze and enhance image details while maintaining the color fidelity and protect privacy. By working in the radiance domain, the system could handle large brightness differences among the color and NIR images and achieve High Dynamic Range (HDR). We proposed two methods to correct the fusion color: linear scalings when raw images were used and color swapping with base-detail image decomposition in the presence of nonlinearity in the ISP pipeline. The system also had two clothing see-through prevention mechanisms to avoid ethical issue arising from the see-through effect of NIR image.

Related Material


[pdf]
[bibtex]
@InProceedings{Ng_2021_ICCV, author = {Ng, Kim C and Shen, Jinglin and Ho, Chiu Man}, title = {A System for Fusing Color and Near-Infrared Images in Radiance Domain}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops}, month = {October}, year = {2021}, pages = {2021-2030} }