Emotion Recognition on Partial Faces with Histogram of Oriented Gradients and Local Binary Patterns

Abstract

Facial expression is a way to improve the ability of communication. If machines can understand the emotion, their responses should be improved. Many researchers can teach the machine to learn emotions from facial images. Sometimes, the face in an image may not be complete, especially when the masks are used in pandemic. Some parts of the human face cannot be reached. However, if the important features can be extracted from the remaining parts, the machine should be able to learn. In this paper, Histogram of Oriented Gradient (HOG) and Local Binary Pattern (LBP) are considered to extract features from the upper-frontal facial images. The concatenation of these features is also studied. The experimental results show the combined features provide the most stability performance with the highest averaged accuracy at 68.39% on Support Vector Machine.