Automatic Hair Detection and Tracking System Using Kinect
Kazumasa Suzuki, Haiyuan Wu
pp. 323-329
DOI:
10.5687/iscie.26.323Abstract
In this paper, we propose automatic hair detection and tracking system at video-rate by using Kinect to capture color and depth information. Our system has the following three ideas simultaneously: 1) Simple and high-speed system is built by the general technique using distance information. 2) Using a 6D feature vector to describe both the 3D color feature and 3D geometric feature of each pixel uniformly. Classifying pixels in images into foreground (e.g. hair) and background with K-means clustering algorithm. 3) Automatic learning and updating the cluster centers of foreground and background before and during hair tracking. This ability makes our system can track hairs robustly, which does not depend on its hair color and style, and even before a background with similar color of hair. Our system becomes a robust head tracking system if the face and hair are set as foreground.