Eye-state analysis using an interdependence and adaptive scale mean shift (IASMS) algorithm

• Published in: Biomedical signal processing and control Vol. 11; pp. 53 - 62
• Main Authors: Ibrahim, Masrullizam Mat, Soraghan, John J., Petropoulakis, Lykourgos
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-05-2014
• Subjects: Adaptive scale; Eye state analysis; Eye tracking; Fatigue assessment; Mean shift algorithm; Mean shift algorithm; Adaptive scale; Eye state analysis; Eye tracking; Fatigue assessment
• ISSN: 1746-8094
• DOI: 10.1016/j.bspc.2014.02.007

•A novel technique integrates a mean shift tracking algorithm with an adaptive scale scheme.•Non-rigid eye movement evaluated from successive frames using eye position on face and distance between the eyes.•The design focuses primarily on aiding human fatigue detection systems.•A specially developed facial fatigue database was used to aid and test the development described in the paper.
Eye state analysis in real-time is a main input source for Fatigue Detection Systems and Human Computer Interaction applications. This paper presents a novel eye state analysis design aimed for human fatigue evaluation systems. The design is based on an interdependence and adaptive scale mean shift (IASMS) algorithm. IASMS uses moment features to track and estimate the iris area in order to quantify the state of the eye. The proposed system is shown to substantially improve non-rigid eye tracking performance, robustness and reliability. For evaluating the design performance an established eye blink database for blink frequency analysis was used. The design performance was further assessed using the newly formed Strathclyde Facial Fatigue (SFF) video footage database11The Strathclyde Facial Fatigue (SFF) video footage database was developed in collaboration with the Psychology Department, University of Strathclyde and the Sleep Centre, University of Glasgow, and it was approved by the Ethics Committee of the University of Strathclyde. of controlled sleep-deprived volunteers.