A Computer Vision System for Analyzing and Interpreting the Cephalo-ocular Behavior of Drivers in a Simulated Driving Context

Abstract

In this paper we introduce a new computer vision framework for the analysis and interpretation of the cephalo-ocular behavior of drivers. We start by detecting the most important facial features, namely the nose tip and the eyes. For that, we introduce a new algorithm for eyes detection and we call upon the cascade of boosted classifiers technique based on Haar-like features for detecting the nose tip. Once those facial features are well identified, we apply the pyramidal Lucas-Kanade method for tracking purposes. Events resulting from those two approaches are combined in order to identify, analyze and interpret the cephalo-ocular behavior of drivers. Experimental results confirm both the robustness and the effectiveness of the proposed framework.