Efficient second-order minimization based visual tracking on moving face

This paper presents a visual feedback control system based on an efficient second-order minimization (ESM) [1] algorithm for tracking a face in motion. This research aims to control the camera installed at the end of a 3 DOF robot arm to fix its orientation to a face in motion so that the moving face appears in a constant position of the image. In this work, the ESM algorithm is firstly derived for feedback of the orientation of the 2D human face. Next, the conventional PD algorithm is applied to the tracking system to track and synchronize with the movement of the face. To track the face and also evaluate the effect of the synchronous motion of the robot system, a webcam is attached in the end effecter of the robot. If the orientation and position of the face of the jogger captured from the webcam appears stationary during its movement, the tracking and synchronous motion system will be proven effective. The experiments verified the effects. This implementation will be further applied on treadmills in which the entertaining LCD panel will synchronously move itself to match the motion of face of a jogger on a treadmill so that the jogger can watch a relatively stationary program on the LCD screen.