Low-light image enhancement based on soft-closing-based illumination estimation and noise mitigation using correlation among RGB components
Low-light image enhancement based on soft-closing-based illumination estimation and noise mitigation using correlation among RGB components
复制标题
DOI：
10.1007/s10043-022-00760-1
复制
发表时间：
2022-09
期刊：
影响因子：
1.2
通讯作者：
Mashiho Mukaida;Seiichi Kojima;N. Suetake
中科院分区：
文献类型：
--
作者：
Mashiho Mukaida;Seiichi Kojima;N. Suetake
到目前为止，已经提出了许多弱光图像增强方法。Guo等人的LIME（IEEE Trans. Image Process. 26（2）：982-993（2017））是基于retinex理论的复杂的低光图像增强方法之一。在retinex理论中，图像被分解为照明和反射。在LIME中，估计照度，并且通过将输入图像除以估计的照度来获得输出图像。LIME可显著增强暗区的对比度。然而，它往往会导致丢失明亮区域中的精细细节和黑暗区域中的明显噪声。本文提出了一种新的弱光图像增强方法来科普这些问题。在该方法中，使用平滑的局部直方图通过软闭合来估计光照。通过将原始噪声抑制方案应用于反射图来获得输出图像。该方法可以提高图像的整体对比度，同时抑制在明亮区域丢失的细节和噪声放大。在实验中，通过与各种先进的微光图像增强方法的比较，验证了该方法的有效性。
Many low-light image enhancement methods have been proposed so far. LIME by Guo et al. (IEEE Trans. Image Process. 26(2):982–993 (2017)) is one of the sophisticated low-light image enhancement methods based on retinex theory. In retinex theory, an image is decomposed into the illumination and the reflectance. In LIME, the illumination is estimated and an output image is obtained by dividing the input image by the estimated illumination. LIME significantly enhances the contrast in dark regions. However, it tends to cause losing fine details in bright regions and noticeable noise in dark regions. In this paper, we propose a new low-light image enhancement method to cope with these problems. In the proposed method, the illumination is estimated by soft-closing using smoothed local histogram. The output image is obtained by applying the original noise mitigation scheme to the reflectance map. The proposed method can improve the contrast of the image overall while suppressing fine details lost in bright regions and noise amplification. In the experiments, the effectiveness of the proposed method is verified by comparing with various state-of-the-art low-light image enhancement methods.