The Research Of Virtual-real Registration Method Of Augmented Reality

In the traditional diagnosis and surgical operation, lesion localization is very important. The accuracy of lesion localization will directly impact the length of operative time, the operation difficulty and final effect. In all types of lesions localization, the intracranial lesion localization is the most complex and difficult. This paper designed an augmented reality experiment system to improve the accuracy of intracranial lesions localization. This system combined with the medical image 3d model reconstruction method and the virtual-real registration technology to make the virtual skull model and the real skull overlapped, so as to realize the intracranial lesions localization and intuitive 3d display. The research of this paper can satisfy the application of intracranial lesion location, it has important research significance and application prospects. Specific research work includes the following contents:1. An intracranial lesion localization augmented reality system was designed and established to accomplish the virtual-real registration of skull model. A registration bracket was designed and fixed on the face of skull model. The bracket was mainly used to analysis the model reconstruction accuracy and assists the virtual-real registration.2. The 3d reconstruction method based on MC algorithm was studied to get the virtual skull model via CT images reconstruction and solve the diagnosis not intuitive problem of 2d medical image. The virtual lesion was added in the reconstructed virtual skull model. The reconstructed virtual skull model was optimized to restore the real structure of real skull and get a vivid display via adding lights and materials.3. The structure of skull model was very complex, so the accurate virtual-real registration of skull model on position and angle was hard to realize only relied on the marker. The registration method combined with marker and registration bracket was proposed in this paper to solve above problem. Through the experiment, the virtual-real registration method proposed this paper was verified via virtual-real fusion display and experimental data analysis.4. All the researches used in this article were validated by experiments. The reconstruction experiment shows that, the CT images reconstruction algorithm used in the paper can accurate restore the structure of real skull. The relative error of relative distance between registration-balls obtained by calculation less than 0.23 mm, it means that the reconstruction accuracy can satisfy the requirement of this paper. The virtual-real registration method proposed in this paper can realize the virtual skull model and the real skull overlapped and virtual-real fusion display. The image point and registration matrix can be used to calculate the space coordinate of corresponding point in the marker coordinates. The average error between real space coordinate and calculated space coordinate is 0.88 mm, conformed to the accuracy demands of virtual-real registration in this paper.