Facial emotion recognition based on Textural pattern and Convolutional Neural Network

Human emotional states can be detected by various ways such as analysingthe brain signals (EEG), speech, skin responses, facial expressions, body gestures etc. But facial expressions recognition is one of the popular techniquesand easy to detect with computers using various computer visions techniques. In this work, we have proposed a new approach to detect basic facial expressions from textural images using deep learning model Convolution neural network (CNN). In the proposed approach textural images of human faces are generated using Local binary pattern(LBP) method and trained the CNN model with the LBPimages. In the proposed approach all the images of extended Cohn-Kanade(CK+)dataset are converted into LBP images that are used to train the CNN model. The performance of proposed facial expression recognition system has been validated on modifiedCK+dataset. Facial emotionRecognition Accuracy of conventional CNN model with CK+ dataset and proposed method on modified CK+ dataset is also compared.