Multi-Source Information Fusion Application To Driving Fatigue Detection

With the rapid development of global economy, the rapid development of automobile industry is speeding up, the automobile quantity increases day by day, while personal safety when driving a car is very important, so the safety performance of the vehicle is also increasing attention. With the rapid development of modern vehicle manufacturing technology, the proportion of safety factors caused by vehicle traffic accidents for increasing is smaller.However, personal factors and the driver has become the main cause of traffic accident. Therefore, the real-time monitoring the fatigue driving state can prevent the safe of driving people is very effective,and can greatly reduce the traffic accident probability caused by driver’s driving fatigue,so as to protect the personal and property safety. Thus,this research has a very important function to social value and economic significance.Firstly, research status, research results and development trend of the domestic and foreign driving fatigue detection technology are studied and summarized. This paper mainly studies the comprehensive judgment of how to detect thefatigue driving based on fatigue characteristicsSecondly, the physiological signal are introduced in this paper, based on the changes of the physiological indexes in the fatigue and non fatigue state, the process and the degree of fatigue can effectively express the ECG and EEG feature value table was initially clear. Based on detailed research on the driver’s facial features, the face recognition algorithm Adaboost algorithm this classic, the face detection method combined with the standard structure of three court five, roughly the area of the eye of a clear positioning, eye detection and then further use of such algorithms are generally range the. Based on the human eye positioning, the eye contour approximation as elliptic mathematical model, according to the long axis and short axis of the ellipse model calculating eyes open percentage, then through the model to calculate the PERCLOS value and the blink time, frequency and other parameters. Analysis of the normal state and the driver fatigue state steering wheel movement characteristics of changes, can judge whether the driver is driving the two main characteristics of fatigue condition--the steering wheel fixed frequency and magnitude of correction.Next, the basic principle of information fusion technology, the functional structure model and other related aspects is studied and summarized, based on the analysis and comparison of common information fusion algorithm, proposed what issues need considered when select fusion algorithms in the practical application.Finally, based on the research of rough set theory, analysis the concrete methods of applicating rough set theory. analysis the discrete normalization and strategy of condition attributes, to establish the quantitative relationship between the single detection characteristics and driver fatigue degree. According to the data reduction, establish the minimal decision algorithm, and judgement of driver fatigue by this decision. Based on multi-source data fusion, according to the particularity of driving fatigue detection, use appropriate information fusion structure, and the rough set theory, tintegrate a variety of fatigue characteristic information, to the judgement of the driving fatigue...