Abstract

Expressions of the human face bring a lot of information, which are a valuable source in the areas of computer vision, remote sensing and affective computing. For years, by analyzing the movement of the skin and facial muscles scientists are trying to create the perfect tool, based on image analysis, allowing the recognition of emotional states of human beings. To create a reliable algorithm, it is necessary to explore and examine a number of factors that might have an impact on the final solution of the problem, starting with the psychological aspects through technological possibilities. The presented experiment used a novel approach to the study of emotional states. Registration was made using high-speed camera Phantom Miro M310 at speed of 1000 frames per second in high definition. This method allows observation of micro changes on the face which represent absolutely new dimension of analyzing emotion. Awareness that most of the research in the field of recognition of emotional states on the basis of the face images, is limited to the analysis only of a single image (represent certain emotional state), by this light the use of high-speed camera and analysis of short video (of a few seconds), consisting of thousands of frames offers great flexibility. The article focuses on the analysis of the correlation of each frame movies from the collected material. Authors presents a hypothesis that the correlation of images might be a tool that allows tracking of changes on the face resulting from stimuli that cause a change in the emotional state. The experiment consisted of recordings of a person's face, when they had to express the emotional state (joy or anger). Tested group consisted of 50 students (32 women and 18 men). Each film has been recorded under an appropriate procedure of processing in order to obtain a uniform research material. The preliminary results give authors the possibility to test the response time to stimuli particular emotional state, evaluation of the muscles, or select a frame on which there is "the greatest intensity of emotions". Taking into account the fact that the analysis of the correlation of the video frame is related to the shape of the human face and the structure of the skin, it is believed that the study can find their application in developing tools for remote detection of emotional states.

Citations

Authors (3)