In this study, an intelligent 3D surface-data extraction system which integrates a Laser Displacement Meter(LDM), a CNC Machining Center(CNC), and a Personal Computer(PC)-486 is developed. In this intelligent system, LDM, CNC, and PC play the roles of distance-measuring sensor, measurement-frame, and main control device, respectively. By enforcing the Direct Numerical Control(DNC)technique, the CNC can be controlled to carry the LDM to measure specified objects or parts automatically. In addition to the effective measurement of thin and complex objects, a new process which combines the contacting and noncontacting methods is established to handle those objects with larger height differences. Furthermore, the Medium Filter technique is applied to filter out the impulse like measuring noise which always occurs in the sharp edges between surfaces. Moreover, a new algorithm is also proposed to efficiently eliminate redundant measured data points. This algorithm enables the users to minimize computing time and reduce memory space during surface reconstruction. Experimental results show that the Medium Filter can successfully wipe out impulse like noise and the proposed redundant data eliminating algorithm can significantly reduce the redundant measured 3D surface data without losing the characteristics of the target surface.