BPSCR: 3D contour-reconstruction method for sheet metal based on binocular photometric stereo vision

3D contour measurement is critical in ensuring proper assembly and optimal performance for sheet metal parts. The contour is a primary feature of sheet metal parts. However, during the stamping process, severe deformation and the formation of reflective roll-over zones often lead to interference factors such as surface discontinuities, viewing angle obstructions, and uneven reflectivity at the contour. Due to these factors, the accuracy and stability of existing methods for 3D contour measurement are insufficient. This article proposes a 3D reconstruction method for sheet metal part contours based on binocular photometric stereo vision. The method enhances images using photometric stereo vision and couples normal and albedo maps for level-set contour extraction. It employs dynamic programming stereo matching based on multiple matching energies, achieving high-precision and high-stability 3D reconstruction of contours with arbitrary shapes. The experimental results show that the proposed method achieves fast and high-precision contour 3D reconstruction, with a measurement accuracy of up to 0.06 mm. Compared to existing structured light methods, it offers higher measurement efficiency and a 73.58% improvement in accuracy; compared to passive binocular 3D reconstruction methods, the accuracy is improved by 53.18%. This method can be integrated with surface-structured light 3D measurement equipment, providing an effective solution for high-precision surface 3D reconstruction and the diverse contour 3D reconstruction requirements.