Recognition of Facial Expression Using Haar Wavelet Transform

This paper investigates the performance of a multiresolution technique and statistical features for facial expression recognition using Haar wavelet transform. Multiresolution was conducted up to fifth level of decomposition. Six statistical features namely variance, standard deviation, mean, power, energy and entropy were derived from the approximation coefficients for each level of decomposition. These statistical features were used as an input to the neural network for classifying 8 facial expressions. Standard deviation from the first level of decomposition was found to give better result compared to other statistical features at different level of decomposition.