Research On An Improved Mass Spring Model And Collision Detection Algorithm

The key problem of virtual surgery simulation is to ensure the accuracy and realtime of soft tissue deformation in the simulation process at the same time.Two commonly used methods of soft tissue modeling are finite element model and mass spring model.Although the finite element method has high precision and good accuracy,the huge amount of calculation results in its inadequate real-time performance,and it is difficult to meet the requirements of real-time interactive feedback for virtual surgery.Mass spring model is easy to calculate,while its accuracy is not high and its shape recovery ability is poor after deformation.An improved particle spring model is proposed to solve these problems,which uses tetrahedron model with vector spring.The restoring force of vector spring and volume force of tetrahedral spring can improve the shape restoring performance and simulation accuracy of the model.The new mass spring model is used to model the soft tissues such as blood vessel,heart and liver.The experimental results show that the model can greatly improve the shape recovery ability of virtual soft tissues,while maintaining good real-time performance.The spring parameters of the existing mass spring model are usually given based on experience,such as spring stiffness coefficient and damping coefficient,which inevitably leads to errors and inaccuracy of the model.In order to obtain accurate model parameters and improve the simulation accuracy of the particle spring model,the stiffness and damping coefficients of the proposed spring model are estimated by using particle swarm optimization.In the process of virtual surgery,the touch and interaction between surgical instruments and soft tissues need to be realized by collision detection algorithm.Considering the real-time requirement of virtual surgery,discrete collision detection is usually used to achieve fast interactive detection.However,the detection time of discrete collision detection algorithm is discrete,which will lead to missed detection.Although continuous collision detection can avoid the problem of missing detection,the computational cost is too high and the real-time performance needs to be improved.An improved continuous collision detection algorithm is proposed to further improve the real-time performance of the mass spring model,which improves the traditional continuous collision detection algorithm from three aspects: model simplification,intersection detection based on motion state classification and initial collision time detection based on dichotomy regression.