Microexpression

Описание: Analysis of Microexpressions in Video Image Sequence Analysis

The analysis of microexpressions within video image sequences involves detecting and interpreting these very brief and subtle involuntary facial expressions, which often reveal hidden emotions. Microexpressions typically last between 1/25 and 1/2 of a second, making them difficult to perceive with the naked eye without specialized training.

The analysis process begins with the detection of microexpressions in the video, often using high-speed cameras that capture a large number of frames per second to avoid missing these rapid movements. Algorithms compare differences between consecutive frames to identify minimal variations in facial muscles, focusing on specific regions associated with emotional expression.

Once key moments, known as the "apex," are detected, the analysis continues by recognizing the type of micro-facial expression (such as surprise, disgust, joy, sadness) from the images around this apex. This recognition frequently relies on machine learning methods based on features extracted from the images, enabling classification of microexpressions with accuracy sometimes exceeding that of untrained human observers.

The main interest of this analysis is to reveal concealed emotions, which is useful in fields such as psychology, neurology, lie detection, and human interaction in professional contexts. However, automatic microexpression analysis remains a technical challenge due to their short duration, low intensity, and individual variations in facial expression.

Thus, microexpression analysis in video sequences combines high-frame-rate image acquisition, precise detection of subtle facial movements, and intelligent classification to decode hidden emotional cues, providing a powerful tool for understanding nonverbal communication.

The program was implemented in 2014 to analyze emotions during patient interviews.