Infrared Image Enhancement Based on Multiple Scale Retinex and Sequential Guided Image Filter

To effectively enhance the visual effect of infrared images, enrich image details, and enhance image interpretability, a multiscale Retinex infrared image enhancement method based on sequential guided image filtering is proposed. Firstly, the original infrared image is decomposed into multiple-scale illumination images and reflection images based on Retinex theory. The illumination images at each scale are processed by low-pass filtering and adaptive gamma correction based on histogram, and the reflection images at each scale are sequentially guided filtered. Finally, the enhanced images at each scale are fused to obtain the final result. Drawing on the idea of sequential processing, a sequential guided filter is designed, in which the enhanced image at the previous scale is used as the guidance image for guided filtering at the next scale, to better achieve contrast, edge, and detail enhancement. Experimental results demonstrate the effectiveness of this method, which outperforms traditional infrared image enhancement methods in both qualitative and quantitative evaluation metrics.