MLSP2011 competition: Face recognition with integrating multiple cues

Automatic face recognition is a challenge task, especially working in practical uncontrolled environment. Over the past two decades, numerous innovative ideas and effective processing approaches have been proposed and developed, e.g. different normalization techniques, intrinsic feature extractions and representation schemes, machine learning methods and recognition mechanisms etc. Processing with integrating multiple cues is an effective approach for upgrading recognition performance. A multiple stage with integrating Harr, gradient and cuvelete features for locating eye center approach is presented. The integration based approach achieved high location accuracy and outperforms other state of the art methods. A face recognition classifier using integrating Gabor feature representation and curvelet feature representation is also presented. The experiment results show that the fusion processing can reduce error rate about 30%, compared with using Gabor feature or curvelet feature alone.