Dual deformable model : real-time high resolution volumetric haptic rendering with realistic surface deformation = 고해상도 물체의 햅틱 렌더링과 사실적 변형이 가능한 실시간 볼륨-표면 이중 가변형 모델
real-time high resolution volumetric haptic rendering with realistic surface deformation
Simulation is an emerging technology and it has many advantages for research, training, and performance assessment. It helps to evaluate the performance of trainees with unbiased assessment and feedback. The experiment with the simulator is performed repeatedly at any time, and the simulator provides the appropriate complexity and difficulty of training according to the current level of trainees. However, developing the useful simulators is very difficult. Especially for the soft object, we should consider the deformations which are computationally expensive. Moreover, if simulators have to support both graphic and haptic rendering and to handle large volume data, implementing the real-time simulators is too hard.
In this thesis, we propose a new model called dual deformable model which is suitable for developing practical simulators with the soft object. Unlike the conventional methods for deformable modeling, we design our new model with the concept that sense of vision is different from sense of touch in real world. Our proposed model employs two internal sub-models: one for visual feedback and the other for haptic feedback. Therefore, we can apply two different algorithms which are optimized for visual feedback and haptic feedback. We are also able to employ surface data or volume data for graphic rendering and haptic rendering at the same time. Although we use two internal sub-models to represent one object, with interaction between graphic model and haptic model, dual deformable model is recognized as one object which can show realistic deformation and utilize internal information.
In the implemented system of dual deformable model, we applied Mass-spring model for graphic rendering and S-chain model for haptic rendering to the virtual liver object extracted from Visible Korean Human (VKH) dataset. And we used surface data for Mass-spring model and volume data for S-chain model to speed up the system because internal information is very important fo...