In augmented reality(AR), camera pose estimation is the most important task. However, the performance of camera pose estimation can be degraded due to the jitter problem caused by noise and unintended motion. Conventional jitter reduction methods such as averaging filters induce a lag problem in causal AR environment. In this paper, we propose a new method to reduce the jitter while resolving the lag problem. To estimate stabilized camera pose, the proposed method uses the modified 3-tab average filter whose input consist of the previously stabilized camera poses and the currently measured camera pose. Experimental results demonstrate that the proposed method can achieve significant quality improvement.