Font Size: a A A

Walking Security Alarm System Based On Smart Phones

Posted on:2019-05-12Degree:MasterType:Thesis
Country:ChinaCandidate:Y GuFull Text:PDF
GTID:2428330545985133Subject:Electronic and communication engineering
Abstract/Summary:
Mobile phone addicts keep using phones with their heads down when walking,ignoring various kinds of potential danger around,which may cause traffic accidents.To solve walking security problems for mobile phone addicts,this paper aims to design a walking security alarm sofware based on smartphone platform.The functions of the software are divided into two modes,namely security warning mode and roadblock detection mode.In security warning mode,this paper proposes a brand new step counting algorithm based on SVM(Signal Vector Magnitude)changing regularity,judging whether the phone user is walking.Meanwhile,we use FaceDetector function provided by Google to judge if there are any faces in the picture taken by the phone's front camera,judging whether the phone user is reading phone screen information.When the user is detected reading phone screen information with their heads down when walking,then the phone will send out warnings.On the other hand,this paper proposes a brand new fall-down detection algorithm based on SVM changing regularity.When the user is confirmed to have fallen down unexpectedly,the phone can send messages to the emergency number for help in time.In roadblock detection mode,we propose an approximate image matching algorithm to judge if the picture taken by the phone's rear camera contains any roadblock information of specific types,with the help of functions provided by OpenCV,such as image filter,egde detection,contour finding and shape matching.Through the form of template matching,we have realized the detection of well covers,poles and zebra crossings inside road images.The test results show that the step counting algorithm and the fall-down detection algorithm have both reached an average accuracy of over 97%,with the operation time both within 0.03 seconds.Simultaneously,the image matching algorithm has reached an average accuracy of 90%,with the operation time within 0.15 seconds.These three algorithms have realized relatively short computing time and relatively high accuracy,on the basis of low system resource occupancy rate,while different parts of the system can operate independently,showing good prospects for application.
Keywords/Search Tags:Mobile phone addicts, Smartphone, Step counting algorithm, Fall-down detection, Image matching
Related items