[4Zoltan]

Mixed Raster Content compression is the answer.

Look at this PDF

http://www.rrc.state.tx.us/rules/comments-3-13-TalismanEnergy.PDF

It has letters that are pixel-for-pixel identical. In the second row the name “Smelley, Ronald”, the “e”s and the “l”s are pixel identical.

It’s how the scanning software creates the smaller file size.

[MMaschin]

Here's a link to how MRC actually works.

http://r.search.yahoo.com/_ylt=A0SO8wdYFuFTsTcA2WBXNyoA;_ylu=X3oDMTEzNWllOW1jBHNlYwNzcgRwb3MDMgRjb2xvA2dxMQR2dGlkA1ZJUDQ4OV8x/RV=2/RE=1407289049/RO=10/RU=http%3a%2f%2fsignal.ece.utexas.edu%2f~queiroz%2fpapers%2fei99mrc.pdf/RK=0/RS=P4xdhsKIX7BdVcagnGqrDOG.Ny8-

First, MRC only applies to multi-layer images - these boxes are contained on a single layer.

The fact that one box is an exact copy of another, can not, and is not a result of MRC. If what you are saying was true, then why isn't the background behind the boxes identical? The answer is because they are separate images contained on separate layers, and therefore there is nothing that MRC will do with them.

To test this fact, simply print off the document and scan it back into a pdf, and see how different the boxes come out to be.