2.5D face landmarking via scale-invariant feature extraction and centroid localization

In this paper, we present and discuss our proposed method on landmarking on 2.5-dimensional (2.5D) face range images. Face landmarking plays an important role as an intermediary component in several face processing operation applications. Locating facial landmarks automatically remains a challenge. Detecting and localizing landmarks from raw face data are often performed manually by trained and experienced scientists or clinicians, and the process is usually lengthy, laborious and tedious. In order to overcome these challenges, we introduce a method that employs geometric approach, through utilizing the mean and Gaussian curvatures, primitive surfaces information to identify and label features as anatomical landmarks. In addition, comparative experiments on both automatic landmarking and manual landmarking were also performed and the results have demonstrated that the proposed method outperforms the manual landmarking in terms of obtaining distinct facial landmarks correctly and accurately.