Related Experiment Video
Updated: Jun 9, 2026
07:03
Medical-grade Sterilizable Target for Fluid-immersed Fetoscope Optical Distortion Calibration
Published on: February 23, 2017
Explainable blind image quality assessment with closed-loop semantic guidance and distortion diagnosis
Chenye Song1, Fujiang Yuan1, Zhiwang Zhang2
1School of Computer Science and Technology, Taiyuan Normal University, Jinzhong, 030619, Shanxi, China.
Scientific Reports
|May 9, 2026
Summary
This study introduces an interpretable blind image quality assessment framework for automated visual systems. It enhances diagnostic support and semantic awareness, improving decision-making in image processing pipelines.
Area of Science:
- Computer Vision
- Image Processing
- Machine Learning
Background:
- Automated visual systems require reliable perceptual inputs for computer vision tasks.
- Current blind image quality assessment (BIQA) methods lack interpretability and diagnostic information.
Purpose of the Study:
- To develop an interpretable BIQA framework for automated visual systems.
- To provide diagnostic support and semantic awareness for image quality assessment.
Main Methods:
- Incorporated semantic priors from a vision-language model into a Vision Transformer.
- Utilized feature-wise linear modulation for content-aware quality evaluation.
- Jointly optimized a distortion diagnosis branch for degradation identification.
Main Results:
- Achieved high consistency with human judgments (Spearman's rank correlation coefficients of 0.9509 on TID2013 and 0.9408 on KADID-10k).
- Demonstrated efficient operation at 65 frames per second.
- Provided structured diagnostic cues for adaptive restoration.
Conclusions:
- The proposed framework offers a balance of accuracy, interpretability, and efficiency.
- Enables content-aware quality evaluation and diagnostic support for automated visual systems.
- Addresses limitations of traditional black-box BIQA methods.
