The development of virtual reality technology promotes the progress of medicine and has a pivotal role in dentistry. This paper focuses on collision detection and force feedback haptic perception for oral virtual surgery. The study uses a combination of octree and AABB enclosing box for collision detection, and a force feedback computation method based on meshless SPH to construct a force feedback model after collision detection for oral virtual surgery. This paper’s octree AABB structure algorithm achieves the shortest time in collision detection experiments, with a 29.82% and 18.03% increase over the traditional AABB algorithm and binary tree structure, respectively. The accuracy of the feedback force in this paper has been improved by 2.18%, which is closer to the operation effect of actual clinical surgery. The model presented in this paper has specific feasibility and accuracy, and can be used for collision detection and force feedback sensing in oral virtual surgery.