Feature Extraction and Localisation on 2.5D Face Range Images

SukTing Pui (Universiti Malaysia Sarawak & SEGi University College, Malaysia); Jacey-Lynn Minoi (Universiti Malaysia Sarawak, Malaysia)

In this paper, we develop an automatic landmarking method on face data using 2.5-dimensional (2.5D) range images. Automatic facial landmarking is a vital process which could be employed in any face applications for analysis, registration and recognition. This process is able to locate face feature points (the eye corners, the nose tip, the mouth corners, chin etc.) without the intervention of human. Automatic landmarking has a number of added advantages over manual landmarking and it is more accurate and less time consuming especially if the dataset is large. We also developed an interface to ease the visualization of the landmarking process. It has interactive tools which allow the manipulation of threshold values. The threshold values are then analyzed and generalized to best detect and extract important keypoints or/and regions of facial features. The results of the automatic extracted facial features and candidate landmarks are shown in this paper.

Journal: International Journal of Simulation- Systems, Science and Technology- IJSSST V15

Published: Jun 30, 2014

DOI: 10.5013/IJSSST.a.15.03.06