Image enhancement, which aims at improving the contrast of low-light-level image and shows the details in the dark, is a widely used technology. When traditional multi-scale retinex (MSR) is used for low-light-level image enhancement, enhancement result will be over-enhanced and the so-called halo will occur. In order to overcome this shortcoming, the result of MSR needs to be further processed. First, MSR result and low-light-level image are fused with weight related to the grey value of the pixels. The result shows that the halo is significantly suppressed, but the brightness of the image is low. Thus, we fuse the image again and take the fusion image which fuses the first step result with the MSR result as the final result. Experimental results show that the proposed algorithm can suppress the halo and preserve the details.

中文翻译：

---

基于fusion和Retinex的微光图像增强

图像增强，旨在提高低光级图像的对比度并在黑暗中显示细节，是一种广泛使用的技术。使用传统的多尺度视网膜（MSR）进行低光级图像增强时，增强结果会被过度增强，产生所谓的光晕。为了克服这个缺点，需要对 MSR 的结果做进一步处理。首先，MSR结果和低光级图像通过与像素灰度值相关的权重进行融合。结果表明光晕被明显抑制，但图像的亮度较低。因此，我们再次融合图像，将第一步结果与MSR结果融合的融合图像作为最终结果。实验结果表明，该算法能够抑制光晕并保留细节。