| According to the statistics reported by Transportation Department, traffic accidents frequently happened in China, the harm and loss which caused is the most serious in the world. Fatigue driving is one of the leading causes, so the research on fatigue driving detection is becoming more and more popular at home and abroad. If it can find a judgment on detecting drivers'fatigue, and make a warning when the driver feels sleepy, it will be meaningful to decrease traffic accidents and increase road safety.Firstly, this paper describes this research background, and refers to a large number of current research statuses and results at home and abroad. After analyzing advantages and disadvantages by synthesis, this paper designs a drivers'fatigue detection system based on ADSP-BF548, and it utilizes eye features fatigue detection algorithm based on PERCLOS—it captures driver's frames in real time by on-board camera and localizes driver's eye position by frame-difference projection, edge extracting and template matching methods, and then it predicts and tracks eye position by Kalman filter algorithm; detects status by combining similarity and gray variance projection methods, lastly it uses value of PERCLOS (the proportion eye closed in a certain time) to detect driver fatigue status, if this value is higher than threshold, the system must make a warning. On hardware selection, it uses ADI ADSP-BF548 hardware platform for algorithm transplantation, which includes video in and out and image storage modules. Some optimization is need on DSP system and code level to ensure this detection system real time performance.The detection system based on image/video processing is regard as the most effective method to solve the fatigue driving problem. It is obviously that research and development on driver fatigue detection system are of high value in the marketing application. |
PDF Full Text Request