Yahoo Pixel Density Filter to Fight Image Spam
From the very beginning of mass email communication, spam has always been a cat and mouse game between email providers and spammers.
Bill Gates may
have predicted the end of spam for 2006, but if you are using a typical email inbox you still rely on a spam filter and good portion of common sense to block certain messages. Images have long been a way for spammers to circumvent spam filters, but Yahoo has recently patented an approach that promises - to a certain degree - spam detection in images.
In a patent that was granted on February 1, Yahoo discusses statistical analysis on "differences in edge pixel distribution patterns." Yahoo believes that there are certain indications that hint to a likely spam image. In that case, the image would be converted to a grayscale image and exposed to edge detection:
"By converting to a grayscale image edge detection may be performed, followed by, in one embodiment, elimination of non-maxima and thresholding of weak edges. Thresholding of weak edges may be performed, in one embodiment, by setting a pixel to black if its chrominance is below a defined threshold value. Edge pixels [are] then employed to determine a pixel density distribution (PDD). A normalized pixel density differential may then be determined from the PDD. Various statistical analyses may then be applied to the resulting normalized PDD to determine a likelihood that the image is spam. If the image is determined statistically to be spam, various actions may be performed to prevent the image from being sent to a destination, including blocking the message, expunging the message of the image, or the like."
Yahoo said that the technology works across any image type, even animated versions. So, how well does the invention word? Yahoo does not reveal any numbers besides thresholds in which an image is determined to be spam or not. Even if it works very well, this might just be another cat and mouse episode as it may not be too difficult to change images in a way so that they do not fall in the range of Yahoo's spam assumptions.
- AT&T May have Inflated iPhone Data Bills
- LinkedIn: The IPO Chain Reactor, Raising Money
- Nokia Phone Gets 1.6 GHz Intel CPU
- AT&T Rolls Out New Data Plan for Heavy Users
- Dumping AT&T? Verizon iPhone 4 Pre-orders Open
- Less Revealing Body Scanner Demoed in Airports
- 'Plenty of Fish' Breach Involves Blackmail, Mafia?
- Side-by-Side Comparison of iPad 2, iPad 1 Screens
- People Returning the Galaxy Tab at a Rate of 16%
- WikiLeaks Nominated for Nobel Peace Prize
- Intel: Use A Smartphone, Live Longer
- The Daily 'Newspaper' is Now Published on iPad
- HTC Pyramid Packs Dual-core 1.2 GHz CPU
- VIDEO: PlayStation Phone's 'Thumb Transplant' Ad
- Reuters: We Have a Witness Who Has Seen iPad 2
- VW's Super Bowl Ad Features Mini Darth Vader
- Asus Has a 'Secret Weapon' Against iPad 2
- iPad 2 Revealed at iOS 4.3 Event Next Week?
- Verizon iPhone Preorders Sell Out in Record Time
Until any convicted (an individual is innocent until proven guilty) spammer is executed, the game will continue.
Like the first executed murderer stopped all others from murdering? Yeah right
A patent on this? You gotta be kidding me.
Like the first executed murderer stopped all others from murdering? Yeah right
I guarantee you there would be a whole lot more murders if the penalty was a light as the penalty for spamming is. You get my point...
It seems like alot of their logic is based on the assumption that a banner, ad, "spam image" has a background that is fairly solid. Wonder how it would do on an image that has alot of different colors and gradients.
It seems like alot of their logic is based on the assumption that a banner, ad, "spam image" has a background that is fairly solid. Wonder how it would do on an image that has alot of different colors and gradients.
Not much, the spammers are going to change their images, then yahoo will have to update the code, and the war against spam will continue.
will it be like the program that was meant to filter pr0n images? And identified a portrait of G W Bush as pr0n? lol..
I'd say it would be about as successful as that.. Couldn't they just run an OCR scan on the image? Most spam images have text on them... Surely any decent OCR algorithm could read it? And then just apply the normal text spam filters as before.. Cheaper, simpler, existing technology... Why waste money on something that wont work properly?
pixel thresholding analysis is done by a person watching the image and answering yes or no to the question "is image a spam?"
i like this news, it shows that yahoo has still a lot of good and talented people left in there. i am a loyal yahoo mail user, keep up the r&d
Not much, the spammers are going to change their images, then yahoo will have to update the code, and the war against spam will continue.
The worst part is that with news stories like this the spammers get advanced notice to "yeah, thats why the email is/will be getting more bouncebacks or blocks!"
Eh...
"So, how well does the invention word?"
What?
So how long until they add this to yahoo messenger and it's chat rooms? Their was actually a time when chat rooms were somewhat popular before the rise of social networks like drop "the" facebook and such.
Sue those who pay for the spam.
"So, how well does the invention word?"What?
It does Word well... in fact, it Excel!
just wait. spammers will find a work around.
I will kick a spammer in da ballz!
You can't prevent spam, though efforts to fight it is much appreciated.