Research And Implementation Of Particle Filter Based Human Face Tracking In Standard Natural Classroom

As a new education mode, E-Learning has been developing rapidlywith the spread of information technology. The Standard NaturalClassroom of E-Learning provides varies of hardware systems that are notavailable in traditional classroom, such as face tracking system, laser penhandwriting system, attention detection system. Face tracking systemfocuses on face tracking for teachers or students, thus capturing theconcept of Human, changing the basic communication mode, promotingintellectualization and enhancing emotional care.Real-time human face tracking is an actively researched topicconcerning probability theory, statistical theory, pattern recognition, it’s arather difficult vision based target tracking technology. Face tracking hasbeen successfully applied to intelligent learning, video conferencing,virtual reality, information security and other fields, and is becoming oneof the most important technologies that change human daily life. Till today,implementing a complete, robust and effective tracking system stillencounters a great many problems because of the diversity and complexityof human faces.This paper studies real-time face tracking algorithm based on theparticle filter frame in the environment of Standard Natural Classroom. Wefirst discuss skin color detection technology, analyze skin samplesdistribution under different color spaces, build threshold-CbCr model.Then Local Binary Pattern (LBP) is discussed to extract texture feature. Inregard to particularity of particle filter adapt to non-linear, non-Gaussiansystems, this paper studies particle filter through motion model, measurement model, resampling strategy and other key issues. A trackingalgorithm fusing skin color and texture information is proposed andeffectiveness of this algorithm is demonstrated through experiment results.