Kinematic target surface sensing based on improved deep optical flow tracking

Measuring the surface geometry of mechanical components while they are in motion—gears rotating, pistons reciprocating—is a key challenge for predictive maintenance and precision assembly. Standard structured light methods require the surface to be stationary during multi-frame fringe projection; any motion introduces inconsistencies that corrupt phase retrieval and 3D reconstruction. This paper reframes the problem as a tracking task: rather than requiring a stationary surface, deep optical flow is used to track individual surface points across structured light frames, and the tracked correspondences are used to warp and register fringe observations to a common reference frame before reconstruction. The optical flow network is improved with kinematic priors—since the target’s motion mode is known (rotation axis, frequency), the flow field is constrained to be physically consistent, reducing drift and improving long-sequence tracking accuracy. The motion-compensated fringes are then processed by a standard phase unwrapping pipeline. Published in Optics Express (2023), the method recovers sub-millimetre surface detail from kinematic targets moving at operational speeds, enabling non-intrusive surface health monitoring without stopping or marking the target.

Problem setting

Sensing the surface geometry of kinematic targets—mechanical components undergoing known constrained motion—is critical for machine health monitoring and precision assembly but is complicated by target motion during structured light illumination. This work proposes to use improved deep optical flow tracking to estimate surface point trajectories across structured light frames, enabling accurate surface reconstruction from moving targets. The deep optical flow network is improved with domain-specific training data and a motion-consistency loss that exploits known kinematic constraints of the target.

In the broader publication record, this work appears in Optics Express, 31(23):39007–39019. 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 4, then recomposed for web display.

Representation and setup. This image is extracted from an embedded PDF image object on page 5, then recomposed for web display.

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

Result comparison. This image is extracted from an embedded PDF image object on page 8, then recomposed for web display.

Additional visual result. This image is extracted from an embedded PDF image object on page 8, then recomposed for web display.

Results and impact

The evaluation reported in Optics Express, 31(23):39007–39019 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 15 candidate image objects from paper.pdf and generated the compressed WebP figures used on this page. The local PDF was also optimized from 2,090,623 bytes to 2,086,418 bytes.