This paper presents an approach about head movement compensation for head mounted eye tracker. Eye tracker is a device for measuring eye movements and estimating POG (Point of gaze). When using eye tracker, head movement will cause the error of POG estimated, so the head movement compensation is one of important topics of eye tracker . In this paper, we use scene camera to detect circular patterns pasted on monitor and estimate monitor's depth and 3D position. When calibrating, get the information of mapping function and calibration plane, when using eye tracker, estimate visual axis by them and calculate POG, the experiment result shows this approach can reduce the error caused by head movement effectively.