Research And Application Of Behavior Detection Based On Smartphone Sensors

Recent advances in mobile terminal technology,wearable technology,mobile internet technology,sensor technology and embedded technology have enabled the developed of smartphone sensors(SSs).SS bridges the virtual world and real world,provides a new interface method between human and environment,and makes us to conveniently and effectively express the real world information from the virtual world information.With the rapid appear of wearable equipment,SSs have been widely used in areas of behavior recognition,behavior analysis,intelligent navigation and high racing game.Behavior recognition is generally referred to analyze and infer users’ behavior after processing the data related to users and collected from sensors,and provides better intelligent services.In this project,two main kinds of behaviors will be studies: driving behavior detection(DBD)and pedestrian safe walking detection(PSWD).For DBD,we propose a driving behavior detection system based on a smartphone’s built-in sensor.Traffic accidents resulting from driving behavior and road conditions are crucial problems for drivers.The causes and countermeasure to traffic accidents have been widely studied by researchers.Whereas several approaches have been proposed to ease these problems,most works entail high computational costs or rigid hardware conditions.To address these challenges,we propose HealthDriving,a smartphone-based system for detecting driving events and road conditions solely with a built-in smartphone acceleration sensor.More specifically,we first collect acceleration data from the acceleration sensor of a smartphone on a vehicle,and then utilize an acceleration reorientation calibration algorithm to convert the obtained acceleration data from the smartphone to acceleration data of the vehicle.Finally,we exploit HealthDriving to detect driving events and road conditions,and evaluate the seriousness of the road conditions and driving events by using an efficient scoring mechanism based on the ISO 2631 standard.The extensive evaluation demonstrates that HealthDriving operates successfully with an ordinary smartphone with a low computational cost compared with other methods,and proves the feasibility and effectiveness of our approach.For PSWD,we propose a pedestrian safe walking detection system based on smartphone sensors.In recent years,pedestrians using smartphones while walking for entertainment has become more and more popular.Waking with a smartphone makes pedestrian focus on the smartphone screen instead of the unsafe surroundings.To avoid stumble,fall down,or even collide with other pedestrians,we propose Walk Well,a safe walking detection system that uses smartphones to provide pedestrians with more safety on watching smartphone while walking.More specifically,we first obtain the moving speed of the pedestrian using the gravity and accelerometer sensors on smartphones.Then,we exploit the front camera based on OpenCV4 Android to detect pedestrian’s face and eyes,and decide whether the pedestrian is watching scree or not based on the pupil’s movement posture in the gray-scale picture.If watching time reaches a threshold we set,WalkWell will give a vibration alert to pedestrian through the smartphone.Finally,we implement WalkWell on an Android smartphone,and evaluate the accuracy of the performance,and the results show that it can prevent the potential danger for pedestrians staring at smartphones for long time.