JBIG2 compression would do NOTHING to the boxes on the document. JBIG2 is a font based compressor. The way it works is by determination the original font and character in an image and replacing the image with character set data. This allows the text to be searchable. That is why you know for a FACT that JBIG2 WAS NOT used in this case, the document in not searchable.
“JBIG2 compression would do NOTHING to the boxes on the document. JBIG2 is a font based compressor.”
NBC claims that it is Mixed Rastor Compression that is involved primarily with the “image segmentation” algorithms and not JBIG2 for the WH LFBC pdf.
I am trying to get up to speed, but the “image segmentation” is being claimed to have captured the first image of a box or “e” (a segment) and then if another similar-enough image (segment) is encountered it is ignored and replaced by a duplicate of the first one. This is NOT OCR but based on an algorithm that tries to ID boundaries of any contiguous shape (segment). It could be a number, letter or blob to the human eye but will be treated the same by Mixed Rastor Compression.
If this is true, then the AP/Applewhite copy (claimed to have been a JPEG made from a photo of a hard copy LFBC, NOT a copier scan), was NOT compressed with “image segmentation” according to NBC. Therefor it is claimed by NBC to NOT have any duplicate images when magnified at the bit-map level. The Applewhite image also doesn’t have the clipping mask, with was added by Preview, according to NBC.