Research Of Tracking And Registration Method Based On A Three-dimensional Marker Of Medical Augmented Reality

Medical field needs more preoperative guidance and intraoperative auxiliary urgently because of the characteristic of high risk and the nonrepeatability,therefore,aimed at the application of augmented reality in medical auxiliary operation,3-D registration technology based on vision was introduced,the virtual model from 3-D visualization was registered with the real scene in real-time in order to display the operation more intuitivly.A regular hexahedron three-dimensional marker was designed to improve the efficiency and rate of virtual-real registration.The marker's design,recognition and position estimation algorithm was mainly focused on,and the marker was used in the virtual-real registration experiment to register the virtual model with the real model.The identification and tracking effect of the marker was proved good in the experiment.The following points are the specific contents of the research:1.Against the three-dimensional characteristics of the real skull model and the single experimental scene under the background of medical auxiliary operation,and the insufficient that two-dimensional marker would lose efficacy suddenly under the limit angle of view,a regular hexahedron three-dimensional marker based on hamming code was designed,and every surface was designed as black and white color fast with different coding.The marker detection and location algorithm was realized by program and ID of marker surfaces was identified.2.Pn P multi-point perspective and POSIT algorithm were used to obtain the camera's position for different identification situations of the three-dimensional marker.In order to identify the graphic of every surface on the 3-D marker as much as possible,the marker's side length was determined 8cm by optimizing the scale of the marker through experiment with different sizes marker.MC algorithm was used to get the virtual model from 3-D reconstruction model of skull CT data and the virtual model was optimized.3.Tracking and registration method was verified by experiment,the camera was calibrated to obtain the internal matrix and average calibration error of the 28 images collected was 0.38 pixels.The virtual model was registered with the real scene and displayed on computer screen,the scaffold and the ball on the real skull model was registered perfectly with the virtual model,ensures the accuracy of the virtual-real registration.The three-dimensional marker system realized the 3-d registration effectively,ensured the quickness of the recognition with the identify speed within 60 ms.In the condition that the scene size is 60cm×50cm×40cm with the real skull model size is 20cm×14cm×14cm,selected marker can make the number percentage of identification surfaces for visual surfaces to be 86.67% to ensure the tracking effectiveness and applied to virtual-real registration successfully.The research achievement provides feasible idea of the research on medical augmented reality.