3D Vision Measurement System for Intelligent Open-Die Forging Center

Abstract

In this thesis, a 3-D vision system is established to conduct the real-time measurement task for intelligent-open-die forging center. This system consists of a CCD camera, a slit alaser projector and a Kawasaki JS 6-axis robot all interfaced to a PC for image processing. The linear structured light technique is employed for data acquisition. An optical filtering algorithm for laser stripe segmentation is developed to locate the stripe peak positions to sub-pixel accuracy. A new calibration method is proposed to determine the mapping correspondence from 2-D image pixels to 3-D reference points. A cylindrical scanning scheme is applied to gather data about the 3-D shape of parts, so that the overlap problem is easily avoided. Finally, the cross-section at any place along the height of a workpiece is reconstructed so as to detect shape error for adaptive control of automated open-die forging. This vision system is low-cost, accurate and reliable. A cylinder shape has been successfully reconstructed and measured with it. The experimental results illustrating its performance are reported and discussed.