The implication of spatial temporal changes on facial micro-expression analysis

• DOI: 10.1007/s11042-019-7434-6
• Publication Date: 2019-03-28T02:02:35Z
• Authors: Walied Merghani; Adrian K. Davison; Moi Hoon Yap
• Keywords: Feature (linguistics); Local Binary Patterns; Frame rate; Interpolation; Benchmark (surveying); Expression (computer science)

Facial micro-expression datasets lack consistency and standardisation, with different research groups using various experimental settings, in particular, where the are varied resolution frame rates. To provide new insights into roles of rate resolution, we conduct an investigation use rates on current benchmark (SMIC CASME II). By Temporal Interpolation Model, subsample SMIC (original is 100 fps) to 50 fps II 200 fps. In addition, settings adjusted three scaling factors: 100% resolution), 75% 50%. Three feature types used test performance these which Local Binary Patterns Orthogonal Planes, 3D Histograms Oriented Gradient Histogram Optical Flow. The results showed that could affect recognition, behave distinctively dependent types. This work provides guidelines for future selecting rate, descriptors micro-expressions recognition.