Vision-Based 2D-to-3D Object Reconstruction and G-code Generation Using Raspberry Pi

Low-cost vision-based object reconstruction has gained significant interest due to advancements in embedded computing and additive manufacturing. This paper presents an end-to-end system for vision-based object reconstruction using a Raspberry Pi and an OV5467 camera module. The proposed pipeline captures two-dimensional images of physical objects, extracts object contours using classical computer vision techniques, converts the contours into scalable vector graphics (SVG), and generates corresponding G-code for fabrication using a 3D printer. In the current implementation, the Z-axis is maintained constant, enabling accurate reconstruction of the object’s geometry in the X–Y plane. Experimental results demonstrate that the generated toolpaths closely match the original object contours. The system provides a scalable foundation for future integration of depth mapping techniques to enable full three-dimensional reconstruction.