Overview
ISO 20462-2:2005 - "Photography - Psychophysical experimental methods for estimating image quality - Part 2: Triplet comparison method" specifies a standardized, subjective psychophysical method for assessing still-image quality of both soft‑copy (displays) and hard‑copy (prints) images. The method is designed to provide precise, repeatable scaling of perceived image differences in units tied to just noticeable differences (JNDs), while keeping observer stress low and enabling evaluation of relatively large sample sets.
Key topics and technical requirements
- Two-step procedure
- Step 1 (Category step): Optional initial sorting of all test images into three categories (e.g., favourable, acceptable, unacceptable) to reduce the sample set to a manageable size for detailed scaling.
- Step 2 (Triplet comparison step): Core method where observers simultaneously view and rate triplets of samples.
- Sample selection guidance
- Recommended maximum number of samples for Step 2 is less than 27 to avoid observer fatigue.
- Valid N values follow N = 6K + 1 or N = 6K + 3, where K is an integer (ensures arrangement where each sample pair is viewed together only once).
- Rating scale
- Triplet comparisons are rated into five ordered categories: 1 (favourable), 2 (acceptable), 3 (just acceptable), 4 (unacceptable), 5 (poor).
- Statistical analysis
- Apply Scheffé’s method (described in the standard) to transform triplet ratings into an interval scale.
- Convert the resulting interval scale into JND units for interpretation and comparison.
- Design goals
- Method provides high scalability, low observer stress, suitability for non‑expert observers, and high repeatability.
- Supporting material
- Annexes include comparisons with paired comparison techniques, combinatorics for triplets, standard portrait images, performance data, and conversion procedures.
Applications and users
ISO 20462-2 is practical for:
- Imaging researchers conducting psychophysical experiments on image quality attributes (sharpness, color appearance, noise, exposure).
- Camera and image‑processing developers evaluating algorithms or exposure series.
- Display and print manufacturers performing cross‑media and color‑appearance comparisons.
- Standards bodies and labs needing a repeatable subjective method for validating objective metrics or color‑management workflows.
Benefits include efficient handling of many samples with precise scaling and reduced observer burden compared with exhaustive paired comparisons.
Related standards
- ISO 20462‑1: Overview of psychophysical elements
- ISO 20462‑3: Quality ruler method
- ISO 12232: (relevant for digital camera exposure/measurement standards)
- CIE and ICC guidelines for color appearance and cross‑media matching
Keywords: ISO 20462-2, triplet comparison method, image quality assessment, psychophysical methods, JND, subjective evaluation, Scheffé’s method, soft copy, hard copy.
