Local Correlation for Noisy Facial Expression Images

Facial expression recognition (FER) can be useful in many areas, for research and application. In this paper, we present a system for automatic recognition of facial expressions using local statistical correlation. This FER system is based on template matching method and automatically can detect the frontal face expressions (anger, disgust, fear, happy, sad and surprise) from both static and sequence images. The templates are created for each expression by averaging the grey-scale images for characterizing each facial expression texture. Then, local statistical correlation is used to recognize the expressions. The noisy facial expression images, which is a critical problem, but rarely addressed in the existing works are considered. Gaussian noise is considered for testing FER system. Experimental results on different databases carry out that the proposed method is efficient for facial expression recognition and also the small computational complexity is needed to implement the algorithm. Furthermore, the recognition accuracy of FER system is robust in presence of noise.