Overview
ISO/TR 14997-2:2022 provides practical guidance for using machine vision to objectively assess surface imperfections of optical elements. It defines how to obtain results equivalent to inspector-based methods in ISO 14997-1 and how to apply the drawing-based grades of ISO 10110-7. The technical report focuses on dark-field detection principles and gives setup guidance to achieve fidelity, repeatability and reproducibility in automated surface-imperfection inspections.
Key topics
- Scope and goal: Objective machine-vision grading and inspection of scratches, digs, sleeks and other surface imperfections to match manual inspector results.
- Evaluation limits: How machine systems relate to human visual limits and how to implement equivalent thresholds and decision algorithms (e.g., concentration and accumulation rules from ISO 10110-7).
- Sample preparation: Cleaning requirements and treatment of adhered particles (counted as imperfections if indistinguishable from digs).
- Illumination configuration (dark field):
- Importance of dark-field setups to detect scattered light from surface defects.
- Recommendations to maximize azimuthal and polar coverage (dome, ring or densely packed LED arrays) to reduce missed directional defects (important for long scratches and sleeks).
- Notes on curved surfaces where ideal polar angle varies across the part.
- Dimensional inspection (IMV D):
- Resolution and accuracy considerations for sizing scratches and digs, using Renard R5 grading series and grade cutoffs referenced in ISO 10110-7.
- Brightness, sensitivity and system performance verification.
- Visibility inspection (IMV V):
- Visibility limits, system resolution and grade assignment based on observed contrast and scattering.
- Procedures for verifying visibility system performance.
- Metrological quality:
- Assessment of repeatability, reproducibility and fidelity of machine vision grading.
- Annex linking measurement precision to classification error.
Applications
ISO/TR 14997-2 is practical for:
- Optical manufacturers and coating shops for automated quality control of lenses, mirrors and assemblies.
- Metrology labs and QA engineers implementing machine-vision inspection systems.
- System integrators and suppliers designing inspection hardware and illumination for dark-field surface checks.
- Procurement and supplier/customer arbitration where objective, repeatable records and statistical reporting are required.
Use cases include automated pass/fail inspection, defect grading to ISO 10110-7, statistical process control, and objective arbitration of disputed surface condition claims.
Related standards
- ISO 14997-1 (inspector-based test methods)
- ISO 10110-7 (drawing specifications for surface imperfections)
- ISO 9802:2022 and ISO 3 (Renard series) are referenced for terminology and grading scales.