This paper presents an innovative bionic prosthetic hand control system and a hybrid control method, which aims to improve the functionality and trajectory bionicity of traditional myoelectric prostheses. The grasping process is divided into two phases: pre-grasping and grasping. Visual perception is incorporated during the pre-grasping phase to identify the target object and autonomously plan a human-like grasping trajectory. In the grasping phase, myoelectric signals initiate the corresponding actions. A two-step correction method (TSCM) for improving grasping trajectories is proposed, which maps the motion of a human hand onto a prosthetic hand. The method reduces the mean average absolute error (MAPE) of the trajectories between the human and prosthetic hands by 3.24%. The actual tripod grasping experiments of the prosthetic hand indicate that the hybrid control method proposed in this paper significantly enhances the bionicity of movement. The mean correlation coefficients of the movement trajectories of the human and prosthetic hands in the x and y directions reached 0.989 and 0.872, respectively. And the mean MAPE value in the x and y directions is 4.28% and 4.80% respectively. The method described in this paper will contribute to achieving more natural and efficient human-machine collaborative control of prosthetic hands.