Abstract: Lane departure accidents on highways cause enormous losses, making lane departure warning systems a hot topic. This paper constructs a vision-based lane departure warning system based on DSP and FPGA computing. The paper presents the system's framework and hardware implementation, and details the transmission, storage, and processing of large amounts of data in the vision system. Finally, based on the constructed system, the feasibility of the system is verified through a series of typical vision processing algorithms and software algorithms applied to lane departure decision-making. Keywords: Lane departure warning; Machine vision; Peripheral device transmission 1. Introduction Driving involves the driver, the vehicle, and the road. Drivers perceive road information through their eyes and other physiological organs, making driving decisions based on their understanding and judgment of this information and their monitoring of the vehicle's condition. However, due to factors such as prolonged driving, carelessness, and blind spots in drivers' perception of road information and vehicle conditions, traffic accidents are increasingly becoming a major safety hazard. This paper studies an active safety system for automobiles that simulates human driving behavior to improve vehicle safety, focusing on the driver, the vehicle, and the road. This system implements active safety technologies such as lane departure warning. Real-time performance has always been a major issue in machine vision-based systems, especially for systems like lane departure warning. The algorithms involved cover the entire visual computing process, from image preprocessing to high-level visual computation. This includes low-level, highly parallelizable algorithms for large datasets such as convolution and grayscale transformation, as well as mid-level feature extraction and pattern recognition algorithms, and high-level visual processing methods to understand the road conditions. The versatility and efficiency of embedded systems have always been a difficult trade-off. The emergence of high-speed DSP processing chips and large-scale reconfigurable chips provides a good solution for constructing a more general-purpose vision processing system that achieves high efficiency for both parallel and serial algorithms at various levels of vision. For details, please click: Implementation of a Machine Vision-Based Lane Departure Warning System