A Design of Mandibular Motion Tracking System Based on Video Processing

In clinical medical research, mandibular motion tracking is an important indicator for the diagnosis of mandibular related diseases. This paper proposes a new mandibular motion tracking system with video processing. Four cameras are used as a multi-vision system to complete the mandibular motion tracking by measuring the checkerboard markers on the face bow, and the tracking system is implemented on the heterogeneous platform of ARM + FPGA. A new checkerboard corner detection algorithm is implemented on FPGA to overcome the challenge brought by the recognition speed improving of the checkerboard mark. While the high-precision 3D coordinate transformation algorithm is implemented on ARM to improve the accuracy of the system. The experimental data show that the system can process images of 1280×1024@91fps in real-time with an accuracy of 60um.