CanControls GmbH/ Aachen

Title of Talk

“Photoplethysmography Imaging – Signal, Model and Cognitive States”

Camera based non-contact measurement of skin blood volume changes provides non-invasive and non-obtrusive opportunities of sensing human states and health. In human-machine interaction and affective computing any cooperative user behavior restricts the user acceptance and convenience. However any kind of natural user interaction yields to subject motion and the interpretation of skin perfusion measurement no longer achieve a high level of accuracy. We present a brief introduction to the topic of motion artifacts in terms of its statistical properties and show how suitable operators yield to a stabilization of the perfusion signal under such nuisance attributes. Finally we highlight the importance of stable physiological signals by demonstrating the link between facial behavior, the nervous system and the corresponding impact on human state prediction tasks.

Speaker Bio 

Christian S. Pilz is holding a German engineering degree in media and computer sciences and a M.Sc. in computer sciences. He spent several years in industrial speech processing research where he published more than a dozen patents. During the last years his research focused on computer vision and pattern recognition tasks. He published several papers about the problem of motion artifacts in Photoplethysmography Imaging and he is participating in industrial and public funded research project.