Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
May 9, 2026ElectronicsOpen Access

Pattern Recognition in Semantic Feature Spaces for Image Colorization Quality Assessment

View Full Paper
Ask AI
Bookmark
Share

Authors

IŽIvana ŽegerSGSonja Grgić

Discussion

Loading...

Member takes

Overview

Randomized trial evaluates colorization quality assessment using TRIPSI framework in images, indicating improvement over existing methods.

Key Points

  • This research aims to improve the quality assessment of colorized images by addressing semantic errors in coloration.
  • Developed the TRIPSI framework combining three deep pre-trained models: TOPIQ, LIQE, and DreamSim.
  • The framework uses rank normalization to standardize scores and ensure comparability across datasets.
  • Evaluated model effectiveness across multiple datasets to align with human visual perception.
  • TRIPSI demonstrated high correlation with human perceptual judgments of colorization quality.
  • The framework effectively captured and evaluated color distortions and saturation artifacts across datasets.
  • Results indicate that semantic pattern modeling significantly enhances quality assessment.

Cite This Study

Žeger et al. (2026) studied this question.

synapsesocial.com/papers/69fecfafb9154b0b82876a15https://doi.org/10.3390/electronics15091969
View Full Paper
Ask AI
Bookmark
Share