Research On Driver Fatigue Detection System Based On DM642

Driver fatigue driving is a major factor that causes frequent traffic accidents frequent. Vehicle, non-contact, real-time driving fatigue testing equipment is the main research directions. The main task of this paper is to design algorithm meet the real-time requirements, transplant and optimize the code, realize fatigue detection system on the TMS320DM642 development platform. In order to meet the requirements of real-time, a fast template matching algorithm is designed to locate the human eye based on the traditional template matching algorithm. What is more, optimize the system algorithm based on detailed study of the architecture TMS320DM642 DSP and the optimization method on the TMS320DM642 development platform. The main research work and conclusions are summarized as follows:1.Human eyes detection and location. The traditional template matching method has low efficiency low speed. This paper sums up a fast template matching algorithm. In the beginning, just a small part points are involved in template matching, and gradually increase the number of points involved in template matching. Through the comparison of correlation coefficients to determine whether to increase the number of matching points or to abandon the current matching position and move to a new position for a new match. When calculating the correlation coefficient, we just calculate with the new points, and then merge it with the original correlation coefficient to get a new correlation coefficient. This greatly reduces the computation of the algorithm. The points involved in calculating the correlation coefficient always distribute on the template uniformly. This ensures the accuracy of method. So the algorithm has high accuracy and high speed, and it can satisfy the request of real time.2.Transplantation and optimization. Because the architecture of TMS320DM642 is different from the computer's CPU, in older to the program efficiency we have to do some changes on the codes in the process of the Transplantation. Optimize the system algorithm based on detailed study of the architecture TMS320DM642 DSP and the optimization method on the TMS320DM642 development platform. For the time-consuming eyes location algorithm, we used packing data, expanding the circle, using online functions and other optimization methods to optimize it, and achieve better results.