Designing And Implementation Of Fatigue Driving Detection System Based On FPGA

In many countries, investigation and analysis of traffic accidents indicated that thecauses are drink driving, fatigue driving, overloading driving and illegal driving and soon. By further analysis of the statistical data, we can find that the fatigue driving is oneof the most important reasons caused traffic accidents. The annual number of trafficaccidents caused by fatigue driving now is no less than the number caused by drunkdriving and speeding driving. In order to reduce the traffic accidents, decrease theglobal economic losses and protect the life safety of people, preventing fatigue drivingwill be an effective means, and the detection of fatigue driving is the key to preventfatigue driving. The purpose of this paper is to research and develop a kind of devicethat can implement the real-time detection of driver’s fatigue state. In the premise of notaffecting the driver’s operation and attention, we detect the driver’s fatigue state by thereal-time monitoring the eyes of the driver and alarm the driver if the result of detectionis fatigue, so as to effectively achieve the goal of avoiding the traffic accidents causedby fatigue driving.This paper designs and develops the real-time fatigue driving detection systembased on FPGA, the system mainly includes three parts: infrared light source system,video collection and displaying system, accomplishments system of image processingalgorithms and fatigue evaluation algorithm. At the beginning of the designing, wechoose the best appropriate strategy–single camera infrared image acquisition based onFPGA to achieve the detection system by designing and proving the variousexperiments that achieved the aims of fatigue detection, which are presented byresearchers at home and abroad. According to the “bright pupil” effect phenomenonemerged when the pupil reflect90%of the incident infrared light with wavelength of850nm, the detection system can collect the video image with gray eyes of more salientinformation. The camera can capture the infrared images with small backgroundinterference by leading the optical filter of850nm peak transmittance into the detectionsystem. The designs and accomplishments of all the functions of the detection systemare based on FPGA platform. Firstly, the system stores the video image into the FPGA’sBlock Ram gathered from the camera, and then sends the stored image to the imageprocessing module to segment the eyes of the driver by using the ping-pang operation,at last sends the segmentation results to the display module and fatigue judge module to display and judge the fatigue state, and decides to alarm the driver or not according tothe fatigue judgment. The image processing module designed in this paper adopts twosegmentation algorithms to finish the segmentation of driver’s bright pupil; onesegmentation method is based on the improved threshold, the other is based onimproved PCNN (Pulse Coupled Neural Network) segmentation method using thenearest neighbor clustering criterion. The fatigue judging module uses the PERCLOSfatigue detection algorithm, and judges the driver is fatigue or not by the PERCLOSvalue computed in this module.This paper achieves the real-time fatigue driving detection system in the FPGAplatform by non-contact image acquisition methods and using the digital imageprocessing technology. The final designed internal modules of the detection system andits experimental results indicate that the detection system is stable operation with highspeed processing, has correct functions, segmentation results and fatigue detectionresults, has good ability of real-time processing and achieves the processing speed of25frames/s, which is the camera’s collecting speed. At the same time, the system has gooduniversality, easy to transplant, high integration, and provides an important referencemodel for other designs in related fields based on FPGA.