Emotion Recognition from Micro-Expressions: Search for the Face and Eyes

摘要

In the paper the problem of the emotion recognition from the micro-expressions is considered. The psychological background of this problem from the modern viewpoint is described in short. There are seven universal emotions which are usually could be recognized. The application fields of the emotion recognition are indicated. The most known databases for the facial micro-expression research are highlighted. The state-of-the-art of the micro-expression spotting and recognition methods is given. One of the main preliminary steps before the emotion recognition is investigated more thoroughly. It is the task of searching for the areas of a face and eyes in the image. Two algorithms implemented to solve this task are compared: the algorithm based on the color intensity estimation and the Viola - Jones algorithm. The results for of these algorithms for images of the people with the different skin colors are represented. The advantages and the drawbacks of the considered algorithms are discussed. The survey and experimental research results are proposed to further use in the emotion recognition pipeline.