An Integrated Two-Stage Framework for Robust Head Pose Estimation

Abstract

Abstract. Subspace analysis has been widely used for head pose estimation. However, such techniques are usually sensitive to data alignment and background noise. In this paper a two-stage approach is proposed to address this issue by combining the subspace analysis together with the topography method. The first stage is based on the subspace analysis of Gabor wavelets responses. Different subspace techniques were compared for better exploring the underlying data structure. Nearest prototype matching using Euclidean distance was used to get the pose estimate. The single pose estimated was relaxed to a subset of poses around it to incorporate certain tolerance to data alignment and background noise. In the second stage, the uncertainty is eliminated by analyzing finer geometrical structure details captured by bunch graphs. This coarse-tofine framework was evaluated with a large data set. We examined 86 poses, with the pan angle spanning from −90 o to 90 o and the tilt angle spanning from −60 o to 45 o . The experimental results indicate that the integrated approach has a remarkably better performance than using subspace analysis alone.

Generalized adaptive view-based appearance model: Integrated framework for monocular head pose estimation

“…Some earlier work in static user-dependent paradigm include nearest-neighbors prototype methods [32,13] and template-based approaches [19]. Vacchetti et al suggested a method to merge online and offline keyframes for stable 3D tracking [28].…”

Monocular head pose estimation using generalized adaptive view-based appearance model

“…Some earlier work in static user-dependent paradigm include nearest-neighbors prototype methods [22,11] and templatebased approaches [23]. Vacchetti et al suggested a method to merge online and offline keyframes for stable 3D tracking [24].…”

Head Pose Estimation for Driver Assistance Systems: A Robust Algorithm and Experimental Evaluation

“…Systems have been proposed that compare these views using normalized cross-correlation [3], mean squared error [4], differences in gradient direction [5], elastic graph matching [6], and distance between subspace projections [7], [8]. These prototype methods are appealing because training requires only positive examples, but their winner-take-all pose estimation can be greatly affected by a single noisy measurement.…”