FACIAL EXPRESSION RECOGNITION BASED ON IMPROVED LBP OPERATOR AND K-MEANS CLUSTERING
Yunfei Wang, Hui Ding, Yi Liu, Yanyan Pan · 2015
For facial expression features extraction and automatic clustering, a method is proposed.This method integrates improved LBP operator and K-means clustering.Firstly, facial features are located by integral and variance projection.Improved LBP operator with changed threshold is figured out.Secondly, K-means clustering produces expression templates.Then K nearest neighbor algorithm recognizes facial expression.Meanwhile, the expression templates will be updated by the result itself, optimizing expression templates dynamically.The algorithm is simple and low computational-complexity.It updates the train results dynamically.Results show that this classification method is accurate.