Why is OCR missing text on low-resolution scans?

OCR often misses text on low-resolution scans because the image no longer contains enough reliable detail for the OCR system to distinguish characters accurately.

Main reasons:

• Blurred character shapes: At low resolution, letters lose sharp edges and fine features like serifs, holes in letters (e.g., a, e, o), and small gaps.
• Merged characters: Adjacent letters may touch or overlap, making one word look like a blob.
• Lost thin strokes: Parts of letters like i, j, punctuation, accents, and small marks can disappear entirely.
• Bad line/word segmentation: OCR first tries to find text lines and individual characters. Low resolution makes that layout analysis less accurate.
• Compression artifacts: If the scan is also heavily compressed, blockiness or noise can confuse the OCR.
• Font size matters: Small text at low DPI may become effectively unreadable even if the page looks okay to a human when zoomed out.

Typical remedies:

• Scan at 300 DPI minimum, often 600 DPI for small text.
• Use grayscale or black-and-white with good contrast.
• Apply de-skewing, denoising, and sharpening before OCR.
• If possible, rescan from the original at a higher resolution.

If you want, I can also explain how OCR engines like Tesseract decide whether a character is “good enough” to recognize.