An Improved Artificial Neural Network Based Emotion Classification System for Expressive Facial Images

摘要

Developing systems and devices that can recognize, interpret, and process human emotions are an interdisciplinary field involving computer science, psychology, and cognitive science. A system has been developed in order to formally categorize the emotions depending on facial expressions. The feature selection is done based on facial action coding system which is basically a contraction or relaxation of one or more face muscles. Our goal is to categorize the facial expression using image into six basic emotional states: Happy, Sad, Anger, Fear, Disgust, and Surprise. Extraction of facial features from eye, mouth, eyebrow, and nose is performed by employing an iterative search algorithm, on the edge information of the localized face region in binary scale. Finally, emotion class assignment is done by applying the extracted blocks as inputs to a feedforward neural network trained by back-propagation algorithm.