A new information-localization strategy for machining large thin-walled parts is presented in this paper. This strategy uses sub-areas and fiducials calibration methods to improve processing precision of large thin-walled parts, which are of poor machinability, low rigidity, and poor deformation coupling. In our experiments, the large thin-walled part is firstly divided into several sub-areas based on the actual state of the workpiece sampled by a binocular vision sensor. And then each divided sub-area is calibrated by the machining benchmark fiducials. Finally, the machining error of each sub-area is automatically compensated based on the measurement results of fiducials in each sub-area by the touch probe. Two verification experiments show that the proposed strategy is feasible and efficient.