A computer vision framework for the analysis and interpretation of the cephalo-ocular behavior of drivers

• Published in: Machine vision and applications Vol. 24; no. 1; pp. 159 - 173
• Main Authors: Metari, S., Prel, F., Moszkowicz, T., Laurendeau, D., Teasdale, N., Beauchemin, S., Simoneau, M.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer-Verlag 2013; Springer; Springer Nature B.V
• Subjects: Applied sciences; Artificial intelligence; Blinds; Communications Engineering; Computer programs; Computer Science; Computer science; control theory; systems; Computer simulation; Computer vision; Device driver programs; Driving; Exact sciences and technology; Ground, air and sea transportation, marine construction; Image Processing and Computer Vision; Mapping; Networks; Original Paper; Pattern Recognition; Pattern recognition. Digital image processing. Computational geometry; Road transportation and traffic; Searching; Software; Software engineering; Vehicle safety; Vision systems; Cephalo-ocular behavior; Haar-like features; Driving simulator; Boosted classifiers; Visual search actions; Pyramidal Lucas–Kanade method; Computer vision; Motor car; Software development; Vehicle driving; Vehicle driver; Experimental result; Visual search; Behavioral analysis; Efficiency; Classification; Sparse representation; Pyramidal Lucas―Kanade method; Robustness; Visibility; Mirror; Simulator; Optical flow
• ISSN: 0932-8092; 1432-1769
• DOI: 10.1007/s00138-011-0381-5

In this paper, we introduce a computer vision system specially designed for the analysis and interpretation of the cephalo-ocular behavior of drivers. The system is composed of both hardware and software components and is described in three steps. The first step is devoted to the description of the driving simulator and the developed software. The second step deals with the identification of the driver’s visual search actions using computer vision. The latter are related to specific driving events such as blind spot checking and rear-view/lateral mirror verification. Based on the simulator’s open module, the third step is concerned with the identification of car/road events (overtaking, crossing an intersection) and the mapping of these events with the driver’s behavior. The proposed system will be used by a kinesiology research group for the evaluation and improvement of driver performances in a safe environment (driving simulator). In addition to the controlled environment, a modified version of the system also deals with real driving contexts (i.e. driving in a real car). Experimental results confirm both the robustness and the effectiveness of the proposed cephalo-ocular analysis framework.