A Region Histogram of Oriented Optical Flow (RHOOF) feature for apex frame spotting in micro-expression

Micro-expressions are the momentary facial expressions that reveal genuine emotional state of people. However, the detection and recognition of micro-expression have been greatly challenging. The apex frame which indicates the most expressive state of a micro-expression will be very helpful for further research on micro-expression. But labeling the apex frame manually is very time-consuming. In this paper, we propose a novel Region Histogram of Oriented Optical Flow (RHOOF) feature to spot the apex frame automatically. First, a set of facial landmarks are detected and then 5 Regions Of Interest (ROIs) are selected from facial region based on the frequency of occurrence of action units. Finally, we extract optical flow fields frame-by-frame and compute HOOF in these ROIs. Experiments are conducted on two ideal spontaneous micro-expression databases, i.e., CASME and CASME II. Improvements of 30.77% and 19.04% are achieved respectively in CASME and CASME II when compared to the BS-RoIs.