Driver Fatigue Driving Recognition Based On Multi Features

In recent years,with the increase of car ownership and the number of drivers,the situation of traffic safety is becoming more and more serious.According to the survey,about 20% of road safety accidents are caused by fatigue driving,so it is particularly important to detect the fatigue status of drivers.However,the current fatigue detection algorithm has low accuracy and single fatigue feature,so it is urgent to study a high accuracy fatigue driving recognition algorithm.In this paper,the current popular fatigue detection algorithm is studied and analyzed.Aiming at the existing problems,a fatigue detection algorithm based on the fusion of multiple fatigue features is proposed.The specific research is as follows1.An improved face location algorithm is proposed.Aiming at the problem that the traditional AdaBoost weak classifier has high classification error rate for difficult samples,the cosine similarity of local sensitive hash method is used to find the nearest neighbor of the sample to be detected;Then the samples to be tested are hashed together by local sensitive hashing to construct a new test set;Then,the classification test is carried out,and the classification accuracy is taken as the dynamic weight coefficient to construct a strong classifier to detect face information;Finally,through the simulation test,it is found that the detection rate of this algorithm is 94% in the daytime image,and the false detection rate is 9.3%,while the detection rate of the night image is 87.3%,and the false detection rate is 11%.Compared with the traditional face location algorithm,the detection performance of the improved algorithm is greatly improved to meet the needs of practical application.2.A fatigue feature detection and extraction algorithm based on multi factor is proposed.The improved local binary fitting algorithm is used to detect the face and eyes,and the mouth region is detected by the method of three Court and five eyes.On the basis of aspect ratio,eye pixel ratio and mouth roundness are added to extract and quantify the fatigue features of eyes and mouth.The experimental results show that compared with the traditional extraction method,the detection rate of multi factor fatigue feature extraction is greatly improved.3.A detection algorithm based on logistic regression model is proposed.Aiming at the single problem of fatigue feature in traditional detection algorithm,four feature phenomena of eyes and mouth in fatigue are fused and judged,and they are brought to the logistic regression model for judgment.The experimental results verify the feasibility of the proposed algorithm.4.Build a real-time fatigue driving recognition prototype system.Based on the previous fatigue detection model,the fuzzy reasoning method is used to build a comprehensive fatigue driving recognition system.According to the multi features of eye and mouth fatigue,the fuzzy rules based on eye and mouth fatigue are introduced,and the fuzzy inference rules for fatigue detection are formulated.The experimental results show that the detection rate of fatigue state is more than 95% through the test of multiple experimental video sequences.Compared with the traditional fatigue detection algorithm based on single feature,the detection rate is more accurate.