This article describes a pose measurement and tracking strategy for an automatic tracking-based scanning system through developing a ball-shaped target. The target structure is introduced. Then, to determine the target pose, a coded marker and accurate center extraction method is proposed using a subpixel and cross-ratio invariance principle. Furthermore, the target coordinate system is established by vision-based target calibration. A confidence-based tracker is proposed, enabling higher tracking confidence and accuracy, and keeping a high tracking speed. The stereo vision estimates the 3-D position and orientation of the moving target in real time. The trajectory is smoothed dynamically based on the Kalman filter. Experiments are implemented to verify the tracking speed, accuracy, and spatial accessibility in robot tracking applications. In addition, the proposed target and strategy have good performance in full-space and fall-orientation tracking.