Dimensions
Are characters large enough in the source?
Check scan and image quality before running OCR
An OCR quality check measures source characteristics that commonly affect recognition, including pixel dimensions, contrast, edge sharpness, skew, and text near the crop boundary. It cannot guarantee accuracy, but it can identify problems worth fixing before a large OCR run.
Your source-quality findings are ready
N/Alocal review signal
0 pages0 words0 characters0 ms
The checks identify measurable source risks. They do not guarantee a recognition result or claim to recover detail missing from the pixels.
Diagnose the pixels before judging the text
The checker reports measurable image properties and connects them to practical actions. The result is a preparation checklist, not an invented universal accuracy score.
- Inspect the native pixel dimensions before upscaling.
- Correct meaningful rotation and perspective at the source.
- Retake blurred or clipped images instead of relying on aggressive filters.
Diagnose OCR source problems before changing the recognizer
Use this checker to diagnose the source rather than repeatedly changing OCR engines without knowing what is wrong. Dimensions, contrast, sharpness, skew, and crop-edge activity explain different failure modes and require different fixes.
Contrast
Do strokes separate from the background?
Sharpness
Do real edges survive capture?
Framing
Are lines upright and safely inside the crop?
Match each finding to a practical source fix
A quality report is useful only when it changes what you do next. Retake problems caused by capture, and use cleanup only for information still present in the pixels.
Inspect at actual size
A document can look clean when zoomed out while individual characters contain too few useful pixels.
Fix capture before filtering
Retake blur, glare, severe perspective, and clipped edges whenever the original is available.
Change one variable
Compare one cleanup setting at a time on the same representative region.
Test the hardest page
A clean cover does not represent small print, tables, faded appendices, or camera-captured inserts.
Keep metrics in context
Contrast and edge measurements describe pixels, but language, typeface, layout, and character size still influence the OCR result.
Use the checker on representative difficult pages
Apply each finding with How to Improve OCR Accuracy Before You Upload.
For fixes tied to each source problem, use the scan preparation and OCR accuracy guides.
Best for
Diagnosing scans, photos, screenshots, and representative PDF pages before OCR.
Not designed for
Certifying image quality, restoring missing detail, or predicting exact OCR accuracy.
Review first
Small text, blur, glare, rotation, perspective, clipping, and inconsistent pages.
OCR source-quality questions
Can a quality score predict OCR accuracy?
No. Language, typeface, layout, and document content also matter. The checks identify source risks rather than guarantee a result. Understand what confidence can and cannot say.
Can sharpening repair motion blur?
No. Sharpening may emphasize surviving edges, but it cannot reconstruct strokes removed during capture. Use the source-quality checklist.