The main way to communicate is through non-verbal expressions, although it could totally bemanipulated by the person to give false expression. Unlike ordinary facial expressions, facial micro-expression has characterized by subtle movement and short duration of appearance which unleashes the true expression beyond the control of the person. Due to the nature of micro-expression which is very brief in time and low in intensity, prevalent methods could not come up with its challenges. One of the well-known dynamic texture descriptors is Local Binary Patterns on Three Orthogonal Planes (LBP-TOP) which mainly lacks in grabbing most vital information. To address this issue in this paper, we propose a novel feature extractor called Cubic-LBP that computes LBP on fifteen introduced planes. We demonstrate the effectiveness of these planes to find the apex frame where maximum facial movements within video sequences have occurred. Moreover, the whole process of spotting the apex frame in this paper is done automatically. Achieving results of apex frame spotting is satisfying on CASME and CASME II databases in comparison with most relevant state-of the-art methods
Vida Esmaeili received her B.S. degree in electrical engineering from the Azad University of Abhar, Iran, in 2015 and the M.S. degree in electrical engineering from the Azad University of Qazvin, Iran, in 2018. Since 2019, she has been working toward a Ph.D. degree in communication at Tabriz University, Iran. Her research interests include the area of image processing, machine learning, pattern recognition, and quantum communication.
Share this page on your timeline