This paper proposes an efficient method to detect and track for debris in Geosynchronous Earth Orbit (GEO) using an image sequence and a voting scheme. The proposed method utilizes several motion features appearing in an image sequence which are obtained from a telescope fixed for the azimuth and elevation angles at a certain frame interval. Basically, linear trajectories of debris are detected using line segments which connect candidates of debris in different image frames. The line segments with image displacements over a limit value are removed using a couple of motion prediction techniques for GEO debris. The motion prediction techniques enable us to accelerate the voting scheme based on statistical features on motion direction and displacement computed from the line segments, and finally to detect trajectories of debris while removing ones of stars and noises. The effectiveness of the proposed method is demonstrated through observation image sequences acquired from the image sensor of the telescope called TAOS in Taiwan.