To improve the processing quality of complex curved surfaces and their features in a five-axis hybrid mechanism, a time-optimal speed planning and precise interpolation method for accurately identifying curvature points is proposed. First, the centripetal method is used to generate path node parameters, and intensive processing is performed to generate speed node parameters. Secondly, a B-spline is used to describe the velocity curve, and the third-order constraint model is transformed into a linear model through pseudo-jerk. Then, the speed node parameters are used as segmentation points to construct the time integral function for each segment path, and an adaptive Simpson’s integration method is used to solve the function. Finally, the obtained discrete point information is input to the motion control card for trajectory processing. The experimental results show that the proposed research model in this paper can quickly converge to the global optimum, effectively reduce the curvature point velocity and quickly recover its neighborhood velocity. The total path time calculated only fluctuates by 3.5 μs, comparable to the accuracy achieved by conventional interpolation methods but with a shorter convergence time. The maximum interpolation errors of position and angle are 0.76×10^-3 mm and 0.9×10^-3° respectively, which are smaller than the encoder resolution and meet the processing requirements. Specifically, at the extreme curvature, the proposed method reduces the interpolation error by 95.37% compared to conventional approaches. Therefore, the speed node parameters generated based on the path curve have better curvature identification, and the total path time calculated by constructing the time integral function for each segment path has higher robustness and efficiency.