QAM modulations over wavelet based OFDM channel for facial expression recognition

Facial expression recognition (FER) has attracted significant interest in the scientific community due to its importance for human centred interfaces. This study compare the effect of different modulation for person-independent facial expression recognition from face images. The image is transmitted based on wavelet based OFDM channel. Then, the higher order local autocorrelation (HLAC) is used for feature extraction and the features are classified using the naive Bayesian (NB) classifier. Six different facial expressions are considered. Experiments carried out on Cohn-Kanade database shows comparable performance between different modulation in received images based on classification accuracy. BER performance is also included to observe when FER is applied to the wavelet based OFDM channel in two different QAM modulations, 4- and 16-QAMs. It is shown that the biorthogonal wavelet family of bior5.5 of 4-QAM outperforms db2 of 4-QAM. When the system uses 16-QAM, the BER result also shows that bior5.5 is better than db2.