When the surface texture of a workpiece is extremely fine, cameras struggle to accurately capture depth information, making conventional machine vision methods insufficient for achieving micrometer-scale three-dimensional surface reconstructions. To overcome this limitation, the study focuses on high-precision 3D reconstruction of the surface morphology of milled workpieces. Given the smoothness of milled surfaces, their susceptibility to overexposure, and the difficulty in extracting depth information from two-dimensional image pixels, the paper proposes a novel method that combines phase-shifting with complementary Gray code to achieve micrometer-level surface reconstruction. The superiority of this method over traditional phase-shifting techniques is demonstrated through comparisons of overall morphology, two-dimensional power spectral density (2D PSD), and the average deviations of sampled surfaces. Results show that the proposed method reduces the average relative error in surface deviations by 25.89% compared to traditional techniques. Furthermore, cross-sectional analyses reveal that the reconstructed point cloud surface more closely mirrors the actual peak-to-valley characteristics of the real surface. Experimental results confirm that this method effectively captures the surface features of milled workpieces, indicating broad potential for applications in precision manufacturing.
Read full abstract