Generative deep-learning-embedded asynchronous structured light for three-dimensional imaging

Synchronising a structured light projector with a high-speed camera is a persistent engineering bottleneck: tight timing requirements impose hardware cost and limit frame rates. Asynchronous structured light removes this constraint by decoupling projector and camera clocks, but temporal misalignment between captured frames corrupts standard phase recovery algorithms. This paper introduces a generative deep learning model that learns to reconstruct coherent phase maps directly from asynchronously captured fringe images. Rather than correcting misalignment as a pre-processing step, the network jointly models the fringe formation process and the uncertainty introduced by timing jitter, producing probabilistically calibrated phase estimates. The framework is validated on a range of moving objects under varied lighting conditions. Published in Advanced Photonics (2024), one of the highest-impact optics journals, the work demonstrates measurement accuracy on par with synchronised systems at two to three times the imaging speed. This result is significant for industrial metrology, autonomous driving, and biomedical imaging where both speed and simplicity are critical.

Problem setting

Asynchronous structured light captures fringe patterns without strict hardware synchronisation between projector and camera, enabling higher frame rates and simpler system design, but introduces temporal misalignment that corrupts phase recovery. This work embeds a generative deep learning model—trained to hallucinate the missing or misaligned fringe phases—directly into the asynchronous structured light pipeline. The generative network learns a conditional distribution over phase maps given the observed asynchronous fringe images, allowing accurate phase recovery from fewer, temporally inconsistent captures.

In the broader publication record, this work appears in Advanced Photonics, 6(4):046004. The visual notes below pair the paper’s original figures with a concise reading of the method, experimental setup, and reported results.

Method and visual evidence

The method follows an optical 3D measurement pipeline: acquire coded images, recover phase or geometric cues, compensate the dominant error source, and reconstruct a reliable 3D result.

The extracted figures below show the sensing setup, algorithmic signal flow, and representative reconstruction or calibration results.

Method overview. This image is extracted from an embedded PDF image object on page 7, then recomposed for web display.

Representation and setup. This image is extracted from an embedded PDF image object on page 9, then recomposed for web display.

Experimental evidence. This image is extracted from an embedded PDF image object on page 9, then recomposed for web display.

Result comparison. This image is extracted from an embedded PDF image object on page 10, then recomposed for web display.

Additional visual result. This image is extracted from an embedded PDF image object on page 11, then recomposed for web display.

Results and impact

The evaluation reported in Advanced Photonics, 6(4):046004 uses the extracted figures above to show the method’s measurement, reconstruction, segmentation, matching, or diagnostic behavior on representative experiments. These visuals are paired with the paper’s quantitative or qualitative analysis to make the workflow easier to inspect from the homepage.

Source handling

I extracted 31 candidate image objects from paper.pdf and generated the compressed WebP figures used on this page. The local PDF was also optimized from 10,394,736 bytes to 10,373,100 bytes.