× The internal search function is temporarily non-functional. The current search engine is no longer viable and we are researching alternatives.
As a stop gap measure, we are using Google's custom search engine service.
If you know of an easy to use, open source, search engine ... please contact support@midrange.com.




That would be cool to use the iseries functionality to generate the images, but a quick search on iseries printing (AFP, InfoPrint Server, etc.) didn't indicate that a character could be rotated. Of course Postscript supports it and it's an addon to InfoPrint Server, per the IBM blurbs, and maybe the "precise character positioning and alignment" feature includes that, I don't know.

However, just wanted to clarify that the rotation that is done in most CAPTCHA's isn't that critical to busting OCR's, just slightly overlapping a couple of the characters will do it. So in that regard, the DDS AFP POSITION keyword I saw I think could definitely do that.

But then getting the spoolfile into some graphics format and clipping the characters out and converting into one of the HTML supported formats if for example it was in TIFF would probably require more effort than just generating the images with that Java program. :)

rd


vhamberg@xxxxxxxxxxx wrote:
I've been thinking about native image generation as this thread continures.

Seems one could generate oddly placed numbers into a spooled file using the

various font characteristics and rotation, then run the result through HPT or

InfoPrint Server to get a graphic image - we have a couple converter products

that could help in the effort, but this is not specifically a vendor plug.

There are also a TIFF library and a PNG library of functions that can be

installed into PASE. Maybe these have some uses in such a process.

Eh?

Vern

-------------- Original message -------------- From: Ralph Daugherty <rdjfc@xxxxxxxxxxxxx>
There was a slashdot thread or two on CAPTCHA's a few weeks ago, but no one really offered anything very helpful about what is going on out there.
Lots of quibling over how certain MSFT entities practice it in a substandard way, but for the most part that's just /. being /. However, when I looked at the example CAPTCHA images, they were trivially straightfoward letters for OCR'ing, relatively lined up and well separated.
Displaying in different colors including pastels really screws OCR up, but it's not necessary. The key is to overlap the characters somewhat with characters tossed and turned.
I agree with the suggestion to just generate these images with random number of characters (from three to five, for example) generated at positions that overlap at least two of the characters and store a set of them on IFS with answers in a file keyed by the file name as suggested (by Nathan I think).
My vague understanding from lots of /. references is an implication that CAPTCHA's are forwarded to very, very low paid people assisting URL spammers (not necessarily worded that way elsewhere, my description) to reply to the CAPTCA's. Given that most spamming attempts come from bot networks of random owned PO's, and that responses are fairly quick, it is onconcievable to me that OCR software algorithms have been downloaded to owned bot PC's or that the CAPTCHA images are forwarded and OCR'd elsewhere.
In any event, as I suggest here to do, most CAPTCHA's are not OCR'able anyway due to overlapping and/or very difficult to separate from background characters.
Nathan's suggestion is really quite simple and the way to go.
rd


As an Amazon Associate we earn from qualifying purchases.

This thread ...

Follow-Ups:
Replies:

Follow On AppleNews
Return to Archive home page | Return to MIDRANGE.COM home page

This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].

Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.