Design And Optimal Implementation Of Driver Assistance System In Vehicle With Far-infrared Sensor

With the improvement of citizen’s life level, vehicles have increased gradually in the road.While traffic accidents occur frequently, And It is a serious threat to people’s life and property safety.According to the thermal imaging principle, the Far-Infrared camera is almost free from the impact of light condition. While,the intelligent Driver Assistance System based on Far-infrared image play a positive role to preventandavoid the traffic accident effectively.The research on Driver Assistance System is paid great attention in domestic recently,and multipleautomobile manufacturers are preparing to install this device. But the technology and products still in development. This paper studies and implements thedriver assistance systemfrom the aspects of practical application.This paper adopts far infrared camera as the video collection device, LCD display as the video playing device, C6000 serial DM6437 released by TI company as the hardware disposal platform and implements the system software based on DSP/BIOS real-time operating system. This paper develops the work from four aspects of improving the recognition rate of pedestrian, reducing false-alarm rate, perfecting the real-time performance and providing the communication port. It specifically includes: developing VPSS video sub system device drive, implementing far infrared camera video acquisition and video testing result outputs; designing and realizing multiple ROI preliminary characteristic classifier; combining SVM classifier based on the multi-dimension characteristics of HOG+LBP to judge on ROI so as to improve the performance of system testing; designing and implementing CAN bus communication module of the system, making it possible to send testing results of the system to automobile controlling device in real time and receiving controlling orders from automobile devices at the same time. Besides, in the process of program implementation, it utilizes DMA direct RAM access mechanism and software flow optimization to improve the real-time performance of the system.In the last, this paper conducts the actual tests on the telecommunication function of the system and the performance of pedestrian detection. In the paper, it utilizes large quantities of infrared night videos with different scenes to test and evaluate the performance of the system. Results demonstrate that the driver assistance system with night vision can satisfy the requirements of actual applications. At present, the system of this paper is ready for the pragmatic phase.