Layerview: 3D Print Inspection and Repair

Built the core MATLAB processing and ICP registration pipeline for fringe-projection inspection and repair of 3D-printed parts.

ME 395, Industry 4.0 · Northwestern University · 2025

I built Layerview’s core processing pipeline to locate defects in 3D-printed parts by comparing measured geometry with CAD, supporting the team’s repair-printing workflow.

What I built:

  • A MATLAB pipeline for baseline correction, point-cloud filtering, and sampling the reference STL.
  • PCA-based initial alignment followed by MATLAB’s point-to-plane iterative closest point (ICP) registration.
  • Geometric error maps and residual analysis to compare the scanned surface with its CAD target.

Tools: MATLAB · point-cloud processing · PCA · ICP · STL geometry

Fringe-projection point cloud and STL reference before and after ICP registration
Measured geometry and the sampled CAD reference before and after ICP registration. Figure from the team's final presentation.

The team’s coin-and-letter repair demonstration reduced the reported RMSE from 0.15 mm to 0.10 mm.

Geometric residual map of the defective printed part
Before repair: reported RMSE of 0.15 mm.
Geometric residual map after the repair demonstration
After repair: reported RMSE of 0.10 mm. The two plots use different color scales.

Team: Natalie Gurganus, Zihan Chen, Lisa Kovacs, Jeremy Lu, and Carter Taylor.

Back to projects