Summary: | 碩士 === 國立成功大學 === 電腦與通信工程研究所 === 98 === The Ashi point is one kind of Acupoint, which is usually selected for treatment of local pain. By appropriately pressing Ashi point, it will effectively relieve someone’s discomfort. In this thesis, we develop the Epipolar based Correspondences Matching from Edge contour and Triangle barycentric (ECMET) algorithm for 3D Ashi point localization of back on the vision-based massage machine. The ECMET algorithm applies epipolar geometry and the edge contour constraints of user’s body to find the correspondence of edge points between two images of back. Based on edge points and their correspondences, corresponding triangle pairs are identified. Then the triangle barycentric coordinate scheme is employed to find the correspondences of identified Ashi points with corresponding triangle pairs. Therefore, the 3D positions of Ashi points can be reconstructed from the correspondences of Ashi points and two camera poses. This method has been realized in a massage system, and the experimental results show that the reconstructed 3D positions of Ashi points obtained by this approach can achieve a higher admission rate than that applied in Traditional Chinese Medicine for all experiments.
|