Font Size: a A A

Research On Facial Expression Recognition In Standard Natural Classroom

Posted on:2010-08-13Degree:MasterType:Thesis
Country:ChinaCandidate:J S LiuFull Text:PDF
GTID:2178360275470230Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
With the rapid development of information technology, E-Learning based on video has been playing a more and more important role in education. One can learn anywhere and anytime because E-Learning breaks the spatial and time limitation. At the same time, E-Learning has better learning experience because of its personality, intelligence and interactivities with the application of computer pattern recognition, data analysis and information retrieval.As the original data of E-Learning based on video, its classroom that is standard natural classroom should be different from traditional classroom, and it should offer more devices and services, such as laser-pen hand-writing system, speech recognition and facial expression recognition system etc.After research about the human face detection, face alignment, facial expression feature selection, expression classification and PTZ camera controlling, this thesis brings forward a facial expression recognition algorithm based on enhanced local binary pattern with wavelet transform and augmented variance ratio. The experiment in JAFFE has shown the validation of the algorithm, and implements a expression recognition system based on the algorithm.The main work done by this thesis includes:1) Research and compare the advantages and disadvantages of popular expression recognition algorithms.2) Research 2D face model and face alignment based on eye location to decrease the influences of pose.3) Enhanced LBP feature based on wavelet transform algorithm is proposed.4) Apply augmented variance ratio to feature selection for real-time computing.5) Experiment in JAFFE and result analysis and compare between enhanced algorithm and original LBP.6) Design and implement facial expression recognition system in standard natural classroom. The devices and software environment is proposed and PTZ camera controlling is researched. Architecture and module are designed and the main C++ class diagram. The work in the thesis is supported by the National 863 Project 2007AA01Z157, and two pending patents and several papers are accepted by key journals in China.
Keywords/Search Tags:standard natural classroom, facial expression recognition, local binary pattern, wavelet transform
PDF Full Text Request
Related items