Reference material for what part of your problem - using the delta value or actually decoding the barcodes?

There was an article many years ago for decoding barcodes in circuit cellar (www.circuitcellar.com).

As for figuring out the delta value, you'll have to figure out what values are noise and what actually is a valid barcode by doing some testing. Actually it may just be easier doing the differentiation in analog rather than the digital domain - you're choice. What you're looking for is the height of the pulse when the sensor goes from light to dark and dark to light. At some point you'll have a threshold so you ignore blips due to noise and only look at the big blips that represent the barcode stripes. If you're lazy, just purchase a reader wand and worry about the decoding software - that's tricky enough as it is!