Research On Feature Contour Extraction Method Of Side Face In Vehicles Based On Videos

As social economy is improving rapidly,cars are gradually becoming the main way for people to travel.However,with the gradual transition from traditional walking to driving,the cars have not only provided a lot of help to people,but also brought a lot of negative effects.All problems have aroused our concern,especially,among which influence of fatigue driving is the most.The prevention of fatigue driving is an enduring research topic.When the front face recognition system is widely developed,the application of the side face recognition is far less mature in contrast with front face recognition.Therefore,this thesis proposes the extraction of the face contour and feature points information of the driver's side face,aiming to extract these contour information of the driver's side face and the contour information of the feature points,and analyzes it comprehensively.The new measurement standard of fatigue can be done subsequently and applied it to the anti-driving fatigue system as to facilitate further research and testing of fatigue driving.The contents of this article are as follows:First of all,the thesis purposes to analyze and summarize the domestic and foreign face profile extraction algorithms,feature parts extraction status and dynamic picture disposal methods.In view of existing problems,it is necessary and feasible to study side face contour and its feature point regions extraction algorithms.After that,the rough extraction of driver's side face is carried out.After comparing various color space and skin color models,this thesis presents an Elliptic Skin Detection Model by YCbCr color space where locates the rough extraction skin area.Finally,in light of the morphology method,the skin color region is analyzed through experiment and the driver's side face skin area is extracted.Thirdly,this thesis introduces the profile extraction function and uses the image subtraction method to obtain the profile of the face.Comparatively,the former has internal and external boundaries,the latter produces discontinuous edge points.After the comparison and analysis,this thesis raises the better boundary tracking method.Next,the feature points of the side face region are extracted,and the outlines of the driver's eyes,mouth and chin are obtained respectively combined with geometric rules of feature point distribution in facial region.Finally,a simple dynamic analysis of the obtained contour map is made to prepare for the subsequent fatigue driving judgment which lays a foundation on following judgement of fatigue driving.At last,firstly,discuss and analyze the experimental results of this thesis.Then,combined with the feature point contour extraction map,a simple dynamic analysis of driver's physiological condition is carried out through the continuous frame of eye distance image,mouth distance image and the change of mandibular feature point trajectory.At last,compared with the algorithm based on YUV-YIQ space side face contour extraction,the advantages of this algorithm are verified comprehensively.