Welcome! Thanks for visiting!
An image can comfortably replace text we rather not expose directly to Google.
If a review requires disclosure and you want to avoid the Wrath of Google, maybe putting “sponsored review” in an image is so much smarter.
Or not.
Google & Image Spam
01:22: “Dreaded image spam. These embedded images not only cost you time, but they hog memory and clock bandwidth. They’re also tough to catch.
But not for us… Optical Character Recognition from Google Book Search helps us block image spam before it gets to you.”
- Google can OCR process images either on-the-fly or in near real-time
- Hiding text in images is now as useless as using document.write(’hide from Google’)
- Using large images to overlap (search engine) visible text is useless
- Could this technique be combined with facial recognition to come up with an auto-flagging feature for dubious sites?
Recommended:
- Making web images (and everything else?) accessible PowerPoint file of the talk at Google, Kirkland.
- Making Web Images Accessible Video of WebInSight talk at Google, by Richard Ladner.
- WebInSight home page
As posted in Google.
You're welcome to join the conversation; add your response. You can track the conversation using the RSS 2.0 feed.
You can also trackback from your own site.
9 Responses to “Google Sees What You Hide in Images”
Trackbacks
-
Gabfire » Links for week 46 Says:
November 26th, 2007 at 9:22 am[…] Google Sees What You Hide in Images […]
-
Links for week 51/07 Says:
December 21st, 2007 at 7:07 am[…] Google Sees What You Hide in Images […]
November 23rd, 2007 at 5:20 pm
Sounds like more fear than practical.
Google will be leaking bandwidth and computing power all over the place, if it tries to scan every image.
When you check the cache of your pages, do the images cached by Google or still on your server?
November 24th, 2007 at 7:22 pm
I don’t see them doing this for regular websites anytime soon, the sheer computing power it’d take would probably offset the benefits, and even if they do, it wouldn’t be that hard to skew the text so its harder to read, most OCRs can’t handle that.
November 24th, 2007 at 9:16 pm
I just tried searching on SEObythesea.com and couldn’t find it but last year I read about a Google patent with analyzing vector points in an image to be able to read what that image is of. It was use a lot of computation to use this same technology for text but it’s very possible.
Just imagine, duplicate content filters for similar imagery, not just textual content.
November 25th, 2007 at 10:30 pm
Anyway,…as they try to fight paid-links this is maybe one thing Google really plans to try/ do.
November 25th, 2007 at 11:12 pm
LOL! Now if only they could get better at cloaking detection
November 26th, 2007 at 8:50 am
The last 10 seconds can easily be labeled as g4y but I won’t do it.
I’d pick MS Outlook over GMail anytime and I do! Host your own email or enjoy the ‘benefits’ and ‘privacy’ of hosted services.
PS: Nice spam check. I vow? God!
November 29th, 2007 at 10:53 pm
Wow…I didn’t even realize that they could do something like that. Looks like they are really serious about getting rid of paid links. But why?