Joint Compression and Restoration of Documents with Bleed-through
Bibliographic record
Abstract
This paper presents research on the digital restoration of scanned two-sided documents suffering from bleed-through, and the joint compression of the original document and its bleed-through corrected version. It is often required to have easy and efficient access to both the original document and the restored version. The method works simultaneously on both the recto and the verso sides, and requires four steps: i) registration, ii) segmentation, iii) inpainting and iv) compression. The first step involves the registration of the recto and the flipped verso so that bleed-though on one side will be aligned with the original information (called foreground) on the other side. An optimization method based on an affine transformation is used for this step. The second step requires segmentation of each side into four regions: ‘foreground only’, ‘background only’, ‘bleed-through only’ and ‘mixed bleed-through and foreground’. Then, the areas identified as ‘bleed-through only’ are replaced with an estimate of the background using an inpainting technique. Finally the two-sided image is compressed for efficient storage and transmission. Each side is first compressed using any standard efficient document compression algorithm such as JPEG 2000. Then the segmentation information identifying the region ‘bleed-through only’ on each side is compressed using a standard bilevel compression algorithm such as JBIG2. The information required to represent the inpainted sections must also be transmitted.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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 source (direct Gemma or distilled Codex), 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".