It is acknowledged that images taken under low-light conditions are easily affected by low visible light and noise, which can cause important image information loss, low signal-to-noise ratio, blurred edges, and poor subjective vision. Related researchers have targeted some solutions are proposed for the above problems, such as histogram equalization and gamma correction, but all have problems such as edge loss and color distortion. Based on the above problems, this paper proposes a low-light image enhancement optimization algorithm based on frame accumulation and multi-scale $Retinex$ joint processing. First, single-channel image frame accumulation filtering is performed on the low-light image, and then the image is jointly enhanced with the optimized multi-scale $Retinex$ algorithm. The experimental results show that the peak signal-to-noise ratio of the image processed by the joint enhancement optimization algorithm used in this article is increased to 51.2041 dB, which is 15.2633 dB higher than the original image and 1.799 dB higher than the image processed by the traditional $MSRCR$ algorithm, structure similarity increased by 0.12, the enhanced image has higher grayscale resolution and signal-to-noise ratio, while retaining more image edges and detailed texture, reducing color distortion to a certain extent, and the generation of aperture artifacts is weakened. It has a high structural similarity to the original image. The overall quality of the image has been improved to a certain extent, and the subjective and objective evaluations are better than traditional algorithms. Finally, the comparison experiment verifies the effectiveness and practicability of the joint enhancement optimization algorithm in this paper to improve the low-light image quality, Which provides a new pre-processing method for future intelligent target detection, and has important research value.

公认的是，在弱光条件下拍摄的图像容易受到低可见光和噪声的影响，这可能会导致重要的图像信息丢失，信噪比低，边缘模糊以及主观视觉差。相关研究人员已经针对上述问题提出了一些解决方案，例如直方图均衡和伽马校正，但是都存在诸如边缘损失和颜色失真之类的问题。针对上述问题，提出了一种基于帧累加和多尺度的弱光图像增强优化算法。$\mathrm{[R}ËŤ\mathrm{一世}ñËX$联合加工。首先，对弱光图像进行单通道图像帧累积滤波，然后以优化的多尺度联合增强图像$\mathrm{[R}ËŤ\mathrm{一世}ñËX$算法。实验结果表明，本文使用的联合增强优化算法处理后的图像的峰值信噪比提高到51.2041 dB，比原始图像高15.2633 dB，比处理后图像高1.799 dB由传统$\mathrm{中号}\mathrm{小号}\mathrm{[R}C\mathrm{[R}$该算法的结构相似度提高了0.12，增强后的图像具有更高的灰度分辨率和信噪比，同时保留了更多的图像边缘和细腻的纹理，在一定程度上减少了色彩失真，并减弱了孔径伪影的产生。它与原始图像具有高度的结构相似性。图像的整体质量得到了一定程度的提高，主观和客观评价均优于传统算法。最后，通过对比实验验证了本文联合增强优化算法提高弱光图像质量的有效性和实用性，为未来的智能目标检测提供了一种新的预处理方法，具有重要的研究价值。