Head pose tracking based on optimizing normalized cross-correlation

This paper proposes an efficient algorithm for head pose tracking. The most preexisting head pose tracking method usually consists of high cost computing algorithm. Therefore, it is difficult to implement head pose tracking in a mobile system with low performance hardware such as smartphone. In this paper, we propose a low cost computing algorithm for fast head pose tracking method. The proposed method consists of two stages: rough tracking and precise tracking stages. Especially, this paper discusses the efficient design of precise tracking stage. According to the experimental results, we demonstrate that our method increases the tracking performance using the proposed coarse-to-fine random sampling algorithm.