Hyperspectral imaging for automatic detection of process contaminants before paint coating

Technology
In development
University
Scouted & summarized by Halo AI

A hyperspectral imaging system developed at the Polytechnic University of Milan to detect and discriminate manufacturing contaminants on metal surfaces. By analyzing spectral signatures, the system identifies organic residuals before paint coating, aiming to improve production quality and support zero-defect manufacturing objectives. It uses machine learning models and push broom camera technology to automatically segment and classify contaminants in industrial settings.

Overview

This hyperspectral imaging (HSI) system, developed by researchers at the Polytechnic University of Milan, is designed to enhance quality control in industrial manufacturing, specifically for paint coating processes. By automatically detecting and identifying residual organic contaminants on metal surfaces—such as washing machine cabinets—before the painting stage, the system supports the goal of zero-defect manufacturing. It works by combining conventional imaging with optical spectroscopy to sample spatially resolved spectral information, allowing for the discrimination of material types based on their unique spectral signatures.

Technical specifications

The system utilizes a Specim FX17 push broom camera, which captures 224 spectral channels in the 900 nm to 1700 nm range. It employs a collimated LED array light source to illuminate samples. The analysis pipeline involves:

  • Foreground extraction: An ensemble model that segments potential contaminants from the metal background.
  • Machine learning classification: Various algorithms, including Support Vector Machine (SVM), Random Forest (RF), and Multi-Layer Perceptron (MLP), are used to classify and discriminate between different types of organic contaminants.
  • Robot-assisted acquisition: A robotic arm is used to facilitate image collection across both flat and structured surfaces.
Technology readiness level

The technology is at TRL 4 (Laboratory validation). Researchers have successfully implemented and tested an acquisition and analysis system in controlled laboratory conditions, replicating contaminants on flat steel sheets and washing machine chassis. While the feasibility of detection and discrimination has been established, the research notes challenges regarding acquisition speed, lighting for structured surfaces, and the need for further investigation into real-time integration and generalized performance in uncontrolled production environments.

Generated Nov 18, 2026. Please verify key information.

About Polytechnic University of Milan

Politecnico di Milano (Polytechnic University of Milan) is a leading public technical university based in Milan, with multiple campuses across Lombardy and a large applied research enterprise. Companies engage via co‑located labs, shared prototyping and testing facilities, and an on‑campus innovation district and startup accelerator. Internships and thesis collaborations make it easy to scope pilots, access talent, and run joint development on site or at partner locations. Research is backed by competitive European and national funding, including European Commission programs such as Horizon Europe, alongside regional support and direct industry contracts. A dedicated technology transfer office manages IP, licensing, spinout formation, and sponsored‑research agreements.

Halo home
Partner smarter. Move faster.
Get new partnering requests
delivered to your inbox.