Face tracking in Kinect

Face detecting and tracking is a famous computer vision technology. It analyzes the images from a webcam or other input devices, and tries to determine the locations and sizes of human faces from these inputs. Some detailed face parts can also be guessed from the given images, including eyes, eyebrows, nose, and the mouth. We can even determine the emotion of a specific face, or the identity of a human from the face tracking results.

The Microsoft Kinect SDK supports face tracking from Version 1.5 onwards. It requires color and depth images from the sensors (or customized sources) as inputs, and returns the position data of the detected head, as well as some important recording points on the face, all of which can be retrieved or used for reconstructing the 3D face mesh in real time.

We will explain how the Microsoft Kinect SDK declares and generates the face mesh at the end of this chapter.