No-Reference Quality Assessment of Blurred Images by Combining Hybrid Metrics
Bibliographic record
Abstract
No reference or blind image quality assessment (NR-IQA) pertains to the challenge of evaluating image visual quality in the absence of a reference image.NR-IQA is necessary for many applications such as medical imaging and surveillance.Consequently, there is a need to devise a novel metric independent of the pristine reference image.The current NR-IQA metrics' performance may be satisfactory for a specific type of blurring but may prove inadequate for other types.This paper focuses on blurred images and presents a novel NR-IQA metric based on restoration schemes and hybrid metrics.Specifically, we utilize a blind restoration technique to address the issue of image blurring.This restoration technique includes three steps: 1) estimating a point spread function (PSF) from the input blurred image, 2) applying a Winner filter to the blurred image to obtain a deblurred image, and 3) convolving the estimated PSF with the deblurred image to produce the reblur image, which is used as a reference image.Furthermore, we utilize the gradient magnitude similarity deviation (GMSD), structure similarity index method (SSIM), peak signal-to-noise ratio (PSNR), and as potent full reference metrics.These metrics are combined to form a viable strategy to enhance the system's performance.The metric under consideration can promptly evaluate an image's quality without necessitating prior learning or training.Compared to existing IQA models, the proposed metric requires no reference, prior learning, or training procedures, making it more convenient and time-efficient.The experimental findings obtained from the analysis of five IQA databases demonstrate that the metric proposed in this study exhibits a level of performance that is on par with the current leading NR-IQA metrics.The comparative results demonstrate that the proposed method outperforms existing NR-IQA methods such as SSEQ, ENIQA, BMPRI, and BLIINDS-II, with Spearman's rank ordered correlation coefficient (SROCC) values higher than 0.87, 0.78, and 0.88 for Gaussian, motion, and out-of-focus blur, respectively.
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 | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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".