<p>This paper presents a Projection-Based Fitting Method (PBFM) designed for the precise identification of eight position-independent geometric errors (PIGEs) in five-axis machine tools. Traditional identification methods, such as the Least Squares Method (LSM), often suffer from numerical instability and high sensitivity to measurement noise due to ill-conditioned kinematic matrices. To address these challenges, the proposed PBFM leverages a two-stage geometric decoupling strategy, where 3D measurement trajectories are projected onto 2D planes to isolate squareness errors and linear offsets through geometric feature fitting. Simulation results demonstrate that the PBFM maintains high identification fidelity even under noise-contaminated conditions, significantly outperforming conventional numerical regularization techniques. Experimental validation on an industrial five-axis machine tool shows an average error reduction of 90% for the C-axis and 77% for the A-axis. The study further discusses the impact of axis travel constraints, revealing that the superior performance of the C-axis is attributed to its full range of motion, which provides more comprehensive geometric data compared to the constrained A-axis. The high efficiency and robustness of the PBFM make it a viable solution for both initial factory calibration and in-situ thermal error compensation, ensuring sustained volumetric accuracy in high-precision manufacturing environments.</p>