ADAS system solutions include camera-based solutions, radar/ LiDAR solutions and sensor fusion. In the early stage of market development, radar/ LiDAR solutions dominated the market due to the mature radar technology and its immunity to weather conditions. However, with the development of ASIC (Application-Specific Integrated Circuit) and the advancement of image processing algorithms, the drawbacks of radar technology have become prominent: it achieves high accuracy in identifying metal obstacles, yet is incapable of detecting non-metallic obstacles such as pedestrians. Moreover, radar technology fails to accurately identify vehicles approaching from the side, and cannot distinguish lane markings, road debris or potholes.

The visual processing technology of cameras can better identify road signs, pedestrians and other information on the road, and calculate the movement trajectories of pedestrians and vehicles through algorithms. Compared with radar technology, camera-based solutions feature lower costs, more comprehensive functions and relatively high accuracy. Camera imaging-based technologies have been gradually adopted by mainstream manufacturers. Considering the limitations of camera pixel resolution on image recognition technology, as well as the degraded performance of cameras in extreme weather such as fog and rain, camera-centric sensor fusion will become the mainstream trend.

The Internet of Vehicles (IoV) architecture consists of three layers from bottom to top: the perception layer, the network layer and the application layer, which undertake the functions of information collection, transmission and processing respectively. Video acquisition and storage (perception layer), as the underlying architecture of IoV, mainly adopts two core technologies: vehicle-mounted DVR and vehicle-mounted IP Camera. A vehicle-mounted DVR, commonly known as a vehicle video recorder, is an integrated application based on digital video compression and storage, 3G wireless transmission technology, with built-in GPS, automobile black box, CANbus, G-SENSOR and other technologies.

A vehicle-mounted IP Camera is based on Digital Signal Processing (DSP) and network technologies. The CMOS image sensor converts optical signals of scenes into electrical signals; after these electrical signals are converted into digital signals, they are transmitted to the DSP memory through a data interface. The DSP completes image compression and encoding, and simultaneously transmits the data stream to a hard disk or other storage devices for preservation. Vehicle-mounted IP Cameras differ from traditional analog systems and DVRs in terms of transmission distance, scalability and cost.

Vehicle-mounted cameras have a broad application scope. Classified by application fields, they are used for driving assistance (dash cams, ADAS and active safety systems), parking assistance (surround view monitoring) and in-vehicle personnel monitoring (facial recognition technology), covering the entire process from vehicle driving to parking. Therefore, cameras are subject to stringent requirements for continuous working time and operating temperature range. Classified by installation positions, vehicle-mounted cameras are further divided into four categories: front-view, rear-view, side-view and in-vehicle monitoring cameras.