Preprocessing and Descriptor Features for Facial Micro-Expression Recognition

Facial micro-expressions contain signi cant information about how people feel, even when they are attempting to conceal their emotions. Previously, very little research has been done to detect and recognize micro-expressions using computer vision methods. In this paper, detection and classi cation of microexpressions from the Spontaneous Micro-Expression database were implemented, following preprocessing and cropping of raw images using Haar features, using local binary patterns on three orthogonal planes (LBP-TOP) and local gray code patterns on three orthogonal planes (LGCP-TOP) as descriptors and support vector machines (SVMs) as detection and recognition algorithms. Results show accuracy comparable to other work.