Images captured by cameras in low-light conditions have low quality and appear dark due to insufficient light exposure, which critically affects the view. Most of the traditional enhancement methods are based on the entire image for exposure enhancement, so overexposed areas in the image have the risk of secondary enhancement. In order to fully consider the exposure in low-light images, we propose a low-light image enhancement based on multi-illumination estimation, which can robustly produce high-quality results for various underexposures. The core of the proposed method is to derive multiple exposure correction images using light estimation. Then, we used a Laplacian multi-scale fusion method to combine the weight map and the images with different degrees of exposure. We used gamma correction and inversion on the original image to produce images with different exposure levels (such as underexposure, overexposure, and partial area overexposure and underexposure). The gamma-corrected image is used for lighting adjustment of underexposed areas in low-light images, while the inversion image is used for adjustment of the overexposed regions. We performed experiments on various images using multiple methods and evaluated and compared the experimental results, qualitatively and quantitatively. Experimental results show that the proposed method in this study can effectively eliminate the effects of low light and improve image quality.

相机在弱光条件下拍摄的图像质量低下，并且由于光线不足而显得暗淡，严重影响了视图。大多数传统的增强方法都是基于整个图像进行曝光增强的，因此图像中曝光过度的区域具有二次增强的风险。为了充分考虑弱光图像的曝光，我们提出了一种基于多重照明估计的弱光图像增强功能，可以针对各种曝光不足情况稳健地产生高质量的结果。所提出方法的核心是使用光估计来得出多个曝光校正图像。然后，我们使用拉普拉斯多尺度融合方法将权重图和具有不同曝光度的图像进行组合。我们对原始图像进行了伽玛校正和反演，以生成具有不同曝光水平（例如曝光不足，过度曝光以及部分区域过度曝光和曝光不足）的图像。伽玛校正后的图像用于调整低光图像中曝光不足区域的亮度，而反转图像用于调整曝光过度区域。我们使用多种方法对各种图像进行了实验，并定性和定量地评估和比较了实验结果。实验结果表明，该方法能够有效消除弱光影响，提高图像质量。伽玛校正后的图像用于调整低光图像中曝光不足区域的亮度，而反转图像用于调整曝光过度区域。我们使用多种方法对各种图像进行了实验，并定性和定量地评估和比较了实验结果。实验结果表明，该方法能够有效消除弱光影响，提高图像质量。伽玛校正后的图像用于调整低光图像中曝光不足区域的亮度，而反转图像用于调整曝光过度区域。我们使用多种方法对各种图像进行了实验，并定性和定量地评估和比较了实验结果。实验结果表明，该方法能够有效消除弱光影响，提高图像质量。