Blind image quality assessment: from heuristic-based to learning-based
Bibliographic record
Abstract
Image quality assessment (IQA) plays an important role in numerous digital image processing applications, including image compression, image transmission, and image restoration, etc. The goal of objective IQA is to develop computational models that can predict image quality in a way being consistent with human perception. Compared with subjective quality evaluations such as psycho-visual tests, objective IQA metrics have the advantages of predicting image quality automatically and effectively in a timely manner. This thesis focuses on a particular type of objective IQA – blind IQA (BIQA), where the developed methods not only achieve objective IQA, but also are able to assess the perceptual quality of digital images without access to their pristine reference counterparts. Firstly, a novel blind image sharpness evaluator is introduced in Chapter 3, which leverages the discrepancy measures of structural degradation. Secondly, a “completely blind” quality assessment metric for gamut-mapped images is designed in Chapter 4, which does not need subjective quality scores during the model training. Thirdly, a general-purpose BIQA method is presented in Chapter 5, which can evaluate the quality of digital images without prior knowledge on the types of distortions. Finally, in Chapter 6, a deep neural network-based general-purpose BIQA method is proposed, which is fully data driven and trained in an end-to-end manner. In summary, four BIQA methods are introduced in this thesis, where the first three are heuristic-based and the last one is learning-based. Unlike heuristics-based ones, the learning-based method does not involves manually engineered feature designs.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.006 |
| Science and technology studies | 0.003 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.004 | 0.002 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".