The Research And Development Of Fatigue Driving Detection Algorithm

Nowadays,road traffic system has experienced rapid development,car share continues to rise. Safe driving has become one of the focal issues of urban development. According to the survey, fatigue driving is one of the major causes of traffic accidents. So, develop a set of real-time fatigue driving detection system that is convenient, efficient and accurate has become the focus of reducing the accident. In the field of image processing, the external characteristic of driver’s eye is one of the external visual features which can directly reflect the fatigue state of the driver’s eye, to design a set of fatigue testing software system that is reasonable and suitable for the promotion. Specifically, in this paper, several key technologies, including face location, face segmentation, eye location and fatigue determination are studied deeply.Specific work of this paper is to design face orientation, face segmentation, positioning and eye fatigue and other major determining core function module algorithm, programming transplantation cross Android platform.Combined with the actual needs of the subject, the experimental images are from the Android smart phone placed in the car, and on this basis to expand the development of the system.Firstly, we improved the existing facial region segmentation algorithm and proposed a face horizontal and vertical segmentation method based on the facial feature information. Vertically, by detecting the position of mouth to achieve a symmetrical division of the face, avoiding the discomfort caused by the simple half the width of the face cutting. Horizontally, we select ears, dark frame of glasses, and other characteristics, to achieve the maximum degree of narrowing of the region of interest(ROI) range and enhance the speed.Secondly, as for the eye detection by wearing glasses, a new method of multi threshold processing is proposed in this paper. In the candidate region of eye, using the ellipse fitting algorithm, the precise location and state analysis of the eye are realized. The experimental results show that the method can achieve 90% recognition accuracy of eye location and state estimation in complex cases, which is better than the existing results.Then, this paper realizes the analysis of the facial orientation based on the improved BP neural network, optimizes the BP network learning algorithm, and realizes the decision of the driver’s head level rotation or the upward movement. With the characteristic information getting from the each chapter before, including the head positioning results and the location of eyes etc, defined the feature vector of the input network, and ultimately achieve the driver’s "distracted" analysis, with the statement of the eye of the driver to make the early warning.Finally, this paper realizes the software porting of cross Android platform, and the algorithm is packaged into APP application, which is successfully applied to the smart phone. By comparing the running results of the mobile phone and the PC machine, the correctness of the transplantation is verified. The results can meet the actual driving needs of people, and be conducive to the popularity of the system.In this paper, we use the C code to realize the function modules of each chapter, using VS2010 integrated development environment development. The final test results show that the system can accurately determine the location of the human eye, state identification and fatigue in a variety of complex situations. The accuracy of the eye location is 93.6%, and the state recognition accuracy is 92.8%. After testing, the system on the PC processing speed is about 15 frames/s, mobile terminal can reach 3 frames/s. The algorithm has high accuracy and good real-time performance, which can meet the actual needs of the daily driving.