Design And Implementation Of Driver's Fatigue Detection System Base On OpenMV

Practical self-driving cars means applied in the field of artificial intelligence in the car are more and more innovative and mature.At the same time,the driving assistance system based on artificial intelligence technologies such as machine vision and voice recognition has attracted much attention of the industry,and the driving safety should be the first consideration in the intelligent and diversified development of automobiles.According to casualty statistics released annual transport sector,fatigue driving has been one of the main reasons for the traffic accident.Therefore,this paper proposes a fatigue driving detection system based on eye feature algorithm.At present,the most commonly used fatigue detection method based on eye features is PERCLOS algorithm,which can judge the fatigue state of a driver by the ratio of the time that the eye closure reaches a certain degree per unit time to the total time.PERCLOS algorithm is presented in the test,a detection algorithm based on blink frequency is proposed,which can detect the fatigue by changing the blink number of the driver per unit time.Besides,the multi-signal monitoring of the driver's eye features,grip strength and the lateral acceleration feature of the car can accurately judge the fatigue of driving from the human physiological signals,the motion feature and the vehicle's motion trajectory.The experimental recognition rate reaches more than 90%.An in-vehicle image processing hardware system with high performance requirements was built at a very low cost.In this paper,OpenMV is selected as the main control of the system to drive the CMOS camera for face image.Eye recognition is realized by Cascade classifier based on Haar feature.After color depth analysis of the eye area,the pupil is recognized,and then the blink frequency of the driver is calculated.And from the driver's blink frequency,pilot grip strength characteristics of the steering wheel and the vehicle's lateral jerkiness three parameters,we can analyze and judge the fatigue of the driver.When the driver is fatigued,the status data will be reported to the cloud server through NB-LOT,and at the same time,the TTS module will be used to continuously remind the driver,until the LD3320 voice recognition circuit receives the specific voice command replied by the driver.In the aspect of data remote processing,the PC monitoring software based on Qt is specially developed for the system,which is used for network acquisition of server data and storage in the local My SQL database for centralized management.The administrator can take necessary early warning measures according to the situation of the driver to prevent further deterioration of the situation.The experimental data obtained after the functional test of the prototype show that the fatigue detection system designed in this paper has realized the fatigue driving judgment based on multiple parameters.Meanwhile,the system has the advantages of low power consumption,low cost,fast detection and intelligent voice interaction,which provides a feasible scheme for the fatigue driving early warning research.