Radon transform aided facial emotion recognition using correlated Hilbert spectrum
Saibal Ghosh, Dwaipayan Choudhury, Arghya Bhattacharya · 2017
This paper proposes an automated system based on Radon's decomposition followed by correlated Hilbert transform for recognizing the emotion reflected on facial images. In this method, signals obtained from successive projection of Radon Transform on facial images of neutral emotion and any other emotion of the same person; are subjected to correlated Hilbert Transform where the correlated pattern of the two first intrinsic mode function (IMF) is obtained. Principal Component Analysis (PCA) by Single Value Decomposition (SVD) and Linear Discriminant Analysis (LDA) are employed successively to attenuate the dimension of feature vectors which are fed to the k-Nearest Neighbor (k-NN), Multi-class Support Vector Machine (MSVM) and Extreme Learning Machines with Radial Basis Function (ELM-RBF) classifiers. On the basis of the experimental studies on “Jaffe” and “Cohn-Kanade” emotion databases, satisfactory efficiency is obtained.