The pixel and geometry kernel behind ScanMate: raster codec, ink separation, warps, homographies, FFT and the pipeline contracts.

The whole ScanMate pipeline: compare a returned document with the one that was issued - the pixel comparison, enhancement, content search and audit, with one class over them all and every other stage loaded only when it is used.

How closely a scan's text matches the original's: OCR matched run by run on the page, figures held exact, every difference located.

PDF pages as rasters with their metadata: text layer with positions, scanned-or-born-digital, real resolution, an original paired with its scan - and where its fields are, found from the labels it prints.

Put a scanned or photographed page back onto the original it came from: deskew, rescale and register, trying transform models until one is good enough.

Merge PDFs, images and rasters into one PDF: PDF pages copied as they are, JPEGs embedded without recompression, every page of a TIFF.

Is a signed PDF still the document that was signed? Verifies each signature's own bytes: the digest, the signature, what it covers, and who signed - without claiming legal validity.