Modern transportation is developing rapidly, bringing with it numerous challenges such as improving traffic efficiency, ensuring traffic safety, and strengthening traffic management. Therefore, finding effective methods for efficient and accurate monitoring of traffic roads is of great significance. By adopting more advanced communication and image processing technologies, it is possible to achieve reasonable control over traffic flow and prevent traffic accidents [1]. Currently, most road monitoring systems in China use wired transmission for information. Although wired transmission offers high bandwidth, its wiring and subsequent maintenance costs are high. Wang Ke et al. [2] used wireless communication to transmit video information. However, due to bandwidth limitations, it could not meet the requirements for real-time transmission, which would reduce the accuracy of road condition assessment. With the application of 5G technology, the speed and stability of wireless communication have been greatly improved [3]. At the same time, adverse weather conditions affecting the clarity of traffic monitoring images also reduce the accuracy of road condition assessment. Wang Weixing et al. [4] proposed an improved Retinex and fractional-order differential-based haze highway traffic image enhancement method, which uses fast guided filtering to estimate the initial illumination component and applies an initial fractional-order differential mask to enhance the reflectance component. Dong Wei et al. [5] proposed an improved Retinex-based urban traffic image enhancement algorithm, which uses guided filtering to obtain the illumination component of the input image, applies a fractional-order integral mask to the reflectance component, and multiplies the processed reflectance component with the illumination component to obtain the final enhanced image. All the above algorithms show good enhancement effects for foggy traffic monitoring images, but in complex outdoor traffic environments, in addition to haze, insufficient light also significantly impacts traffic monitoring image recognition. Although infrared cameras have been widely used in surveillance [6], their use is subject to conditions; only when the object's temperature is higher than the ambient temperature can infrared cameras capture the object. Traditional video enhancement techniques, such as histogram equalization, do not perform ideally in real-time processing of video sequences. To address the above issues, a traffic monitoring system is proposed that utilizes 5G wireless communication for video information transmission. On the server side, an improved single-scale Retinex algorithm is employed to obtain defogged images. The brightness of low-illumination images is then enhanced by inverting the image and utilizing an atmospheric scattering model. Finally, the defogged image and the brightness-enhanced image are weighted and fused to reduce the impact of haze and low illumination on system imaging and improve the accuracy of recognition algorithms.

1 System Overall Scheme

This system consists of a charge-coupled device (CCD) image sensor, a 5G wireless communication module, a server, and a client. The CCD collects traffic video information, and the 5G communication module transmits the collected video information to the service control center. On the server side, video images are first preprocessed. Defogging and brightness enhancement operations are performed on foggy and low-illumination images to improve contrast and clarity, and then stored. Subsequently, a traffic event recognition algorithm is used to identify traffic events. Finally, the recognized information is sent to the client for early warning. The overall system structure is shown in Figure 1.

![](https://pub-048dcb96257f476697b113fcb5939cb9.r2