Digital Filtering of Pupil Diameter Variations for the Detection of Stress in Computer Users

There is increasing interest in developing means to estimate, in real-time, the level of stress of computer users, particularly for applications such as computer-based tutoring. This real-time stress recognition has been attempted through the processing of a variety of biosignals measured from the computer user, such as the Galvanic Skin Response (GSR), the Blood Volume Pulse (BVP), etc. Recent reports in the literature have strengthen the notion that the Pupil Diameter (PD) can also play an important role in the affective assessment of stress in humans, and its differentiation from a baseline state of relaxation. This paper studies different approaches to perform digital pre-processing of the raw PD data towards the detection of stress states in the computer user. The pre-processing consists of removal, by filtering, of abrupt changes in the PD signal that are not likely to correspond to actual pupillary reactions. This study also summarizes preliminary results for stress detection obtained by imposing a threshold on the filtered PD signal.