The Research And Implementation On Fatigue Driving Detection System Based On DaVinci Platform

As the social urbanization and the pace of life speed up, automobile industry isdeveloping rapidly, which leads to an increasing tendency of traffic congestion andaccidents. It can be concluded from statistics that the fatigue of driving holds a very largeproportion, thus making fatigue driving detection, which is used to improve the detectionaccuracy of fatigue state, the key technology that’s to be resolved. What’s more, it hasalso been a hot spot in automotive products.So it has high theoretical researchsignificance and practical application value.In this paper, the fatigue driving detection algorithm is based on the Highperformance digital video development platform--DaVinci. At first, an image processingalgorithm is used to make analytical processing on each frame image.Then we detect theface from the image that has been processed and locate the eyes and mouth in the face.After that,we get to know the drivers’ fatigue state according to the detection of theirmouse and eyes. At last, it can realize real-time processing through complexityoptimization for320x240resolution images.In this paper, along with the car environment and the driver’s behavior patterns, westudy and analyze the characteristics of the human face and make feature extraction andstate detection, then port and optimize the algorithm in view of the DaVinci developmentboard. The core content of research mainly includes：Firstly,we put forward simple andfast face detection algorithm. Secondly, we detect human eyes and realize the accuratedetection of positioning of man’s eyes. Then we realize the positioning of man’s mouseby using Fisher transformation to make the segmentation of lip color and skin color, andthe binarization of gray images.In addition, by extracting the eye and mouth area ofinterest and use hybrid projection of its shape feature to make analysis for staterecognition algorithm,we design a fusion of the fatigue status of judgment method according to the fatigue state decision algorithm of authority and the eyes and mouthfeature information we get. At last, the algorithm is transplanted to the DaVinci platform.The performance of its hardware modules are analyzed and the video data we get isprocessed. Then we studied the DaVinci hardware optimization and optimization ofalgorithms and code. On completion of transplantation of fatigue detection algorithm, wemade system simulation test and verification. Finally, the driver fatigue detectionalgorithm based on the DaVinci platform is implemented and we realized the Driverfatigue detection system.