Facial Expression Recognition Based-On Saliency Guided Support Vector Machine

Facial expression recognition in complex environment is one of the difficult tasks of visual recognition in recent years. This paper introduces the visual saliency mechanism and we design automatic searching of the face region in the image. Using the narrow band C-V model to evolve curve, the proposed scheme can obtain the accurate face region. Meanwhile, the SVM will be trained by standard database. It will calculate the Bhattacharyya distance between the clustering center and the face sample which is extracted from complex environment and then updating the parameters of SVM. Finally, the performance of the classifier will be obtained through testing the test set. Experimental results on lighting change or noisy added image in complex environment confirm the validity of our approach.