Material Vision Lab — Perseus Research
Research ProgramsUltravisor
Experimental System — Material Vision Lab

Ultravisor

Perseus explores how hyperspectral imaging systems could transform material inspection in leather manufacturing. Seeing what the human eye cannot.

Research Initiative — Not a commercial product
The Problem — 01
Leather quality is invisible to the human eye.

The leather manufacturing sector relies on manual classification systems that are subjective, inconsistent, and unable to detect microscopic surface variations. Quality is highly variable. Defects are difficult to see. Grading depends on individual experience and lighting conditions.

The result: inconsistent quality control, material waste, and production delays. Perseus is exploring whether spectral imaging technology can provide an objective, scalable solution.

The Technology — 02

Hyperspectral
Imaging Systems

Hyperspectral cameras capture hundreds of wavelength bands simultaneously — far beyond what the human eye perceives. Each material surface generates a unique spectral signature, a fingerprint of its molecular composition, texture, and structure.

01
Spectral Capture

Hyperspectral camera scans the leather surface across hundreds of wavelength bands simultaneously.

02
Fingerprint Extraction

Each scan generates a unique spectral signature encoding the material's molecular composition and surface structure.

03
AI Analysis

Machine learning models compare spectral signatures against a reference database to detect anomalies and classify quality.

04
Database Archive

Results are stored in a spectral material database, building an intelligent reference library over time.

Potential Applications — 03

What Ultravisor
Could Enable

Defect Detection

Automated identification of surface defects, scratches, grain irregularities and structural anomalies that are invisible under standard lighting conditions.

Potential reduction in manual inspection time
Material Grading

Objective, consistent quality grading based on spectral data rather than subjective visual assessment. Eliminates inter-operator variability.

Consistent grading across production batches
Automated Quality Control

Integration into production lines for real-time quality monitoring. AI models flag non-conforming materials before they enter the manufacturing process.

Real-time inline inspection capability
Spectral Material Archive

A growing database of spectral fingerprints for different leather types, tanneries, and treatments. A long-term infrastructure for intelligent material management.

Foundation for future AI training datasets
UV
Long-Term Vision — 04

A Spectral Database
for Intelligent
Manufacturing

The long-term goal of the Ultravisor research program is to create a spectral material database — a comprehensive archive of material fingerprints capable of supporting intelligent inspection systems across the leather manufacturing supply chain.

This is not just storytelling. The leather sector has real, unsolved problems in material classification. A spectral fingerprint system could become a genuinely transformative infrastructure. Perseus is exploring whether this is technically and economically viable.