High-precision dynamic 3-D reconstruction technology is highly valuable in industrial automation applications, including defect detection, visual measurement, and grasp planning. Presently, predominant 3-D reconstruction for industrial applications requires projecting a pattern sequence, introducing complexities to positioning and measurement in dynamic conditions. We proposed a single-shot dense 3-D reconstruction framework for industrial tasks in response. This work is valuable in the following two aspects. 1) We proposed a design scheme of structured light patterns and developed a multifrequency phase decoding network to extract the wrapped phase from the multifrequency fringe pattern. These unique phase combinations can be used as a robust feature for disparity optimization and stereo matching. 2) We created a dataset with 3000 scenes and their high-precision wrapped phase ground truth. Experiments demonstrate that our proposed method significantly outperforms existing single-shot structured light reconstruction methods with high precision and adaptability. Thus, this framework has significant potential application in industrial automation.
No takes yet. Share an insight, caveat, or question.
Wu et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: