Abstract | ||
---|---|---|
Micro-expressions are facial expressions that have a short duration (generally less than 0.5 s), involuntary appearance and low intensity of movement. They are regarded as unique cues revealing the hidden emotions of an individual. Although methods for the spotting and recognition of general facial expressions have been investigated, little progress has been made in the automatic spotting and recognition of micro-expressions. In this paper, we proposed the Main Directional Maximal Difference (MDMD) analysis for micro-expression spotting. MDMD uses the magnitude of maximal difference in the main direction of optical flow as a feature to spot facial movements, including micro-expressions. Based on block-structured facial regions, MDMD obtains more accurate features of the movement of expressions for automatically spotting microexpressions and macro-expressions from videos. This method obtains both the temporal and spatial locations of facial movements. The evaluation was performed on two spontaneous databases (CAS(ME) 2 and CASME) containing micro-expressions and macro-expressions. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1007/978-3-319-54427-4_33 | COMPUTER VISION - ACCV 2016 WORKSHOPS, PT II |
Field | DocType | Volume |
Computer vision,Expression (mathematics),Pattern recognition,Computer science,Facial expression,Artificial intelligence,Optical flow,Spotting | Conference | 10117 |
ISSN | Citations | PageRank |
0302-9743 | 1 | 0.34 |
References | Authors | |
0 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Sujing Wang | 1 | 690 | 37.65 |
Shuhang Wu | 2 | 1 | 0.34 |
Xiaolan Fu | 3 | 786 | 60.72 |