Single-Shot Three-Dimensional Measurement by Fringe Analysis Network

Fringe projection profilometry (FPP) has been broadly applied in three-dimensional (3D) measurements, but the existing multi-shot methods, which mostly utilize phase-shifting techniques, are heavily affected by disturbance of vibration and cannot be used dynamic scenes. In this work, a single-shot 3D measurement method using deep neural network named Analysis Network (FrANet) is proposed. The FrANet composed phase retrieval subnetwork, unwrapping refinement subnetwork. combination multiple subnetworks can help to recover long-range information that missing for single U-Net. A two-stage training strategy pre-trained fringe pattern reprojection fine-tuned ground truth maps designed. Such lowers number data set, saves time during collection, maintains accuracy supervised methods real-world setups. Experimental studies were carried out on setup FPP system. test mean absolute error (MAE) refined was 0.0114 rad, root square (RMSE) reconstruction results 0.67 mm. proposed scenes evaluated measuring moving standard spheres. sphere diameter maintained high 84 μm at speed 0.759 m/s. Two-stage only requires 8800 images acquisition, while require 96,000 same iterations. Ablation verified effectiveness two stages three subnetworks. achieved accurate measurements comparable those obtained efficiency. This enables shape or vibrating objects industrial manufacturing allows further exploration architecture with few samples measurement.