| Summary: | 碩士 === 長庚大學 === 電機工程學系 === 100 === This paper presents a marker-less medical augmented reality (AR) system, which is applied for craniofacial application. Through a head mounted display (HMD) of the proposed system, the user can observe the patient augmented with by his/her medical images simultaneously.
This system is separated into two stages: preoperative localization stage and display stage. In the preoperative localization stage, we first applied KINECT device to reconstruct a 3-D model of the patient’s head. Second, the CT data of the patient are registered to his/her 3-D model by using WA-ICP algorithm, which was developed by our laboratory previously. In the display stage, the HMD, which is attached with a CCD camera, is prepared to the AR visualization. While the head is detected in the captured image, a feature matching process is applied to align the head and the head model, which is reconstructed in the preoperative localization stage. After the feature matching process, the medical images can be showed on the camera images. In addition, since the AR camera is movable, a feature-point tracking approach which combines KLT tracking and RANSAC-based estimation is employed to provide a stable extrinsic-parameter estimation of the camera. This performance and accuracy of the proposed system are evaluated by using a plastic head and tested with a real person.
|
|---|
