Barracuda Networks Image Spam Protection

Image spam, which generally embeds text inside images with the intent of hiding content from traditional spam filters, represents approximately one-third of all traffic on the Internet. The Barracuda Spam Firewall was the first major email security product to include image analysis techniques to effectively protect against new image spam variants.

These industry-leading techniques include optical character recognition (OCR), image processing and animated GIF analysis. Barracuda Networks image spam defense also calls upon third-generation technology including a multi-pass OCR engine that renders spammers’ tricks to hide text behind color or blurred images ineffective. In addition, it offers automated fingerprint generation for the fastest possible detection of previously issued image spam. These technologies combined garner the Barracuda Spam Firewall an accuracy rating of 95 percent in identifying and blocking image spam. Image spam and fingerprint updates are delivered automatically to all Barracuda Spam Firewalls through Barracuda Energize Updates.

Optical Character Recognition:

Using optical character recognition (OCR), the Barracuda Spam Firewall can identify text hidden within images, convert it to data and then give it a score based on specific rule sets developed by engineers at Barracuda Central. The OCR score is then combined with other factors about how the message is constructed to determine whether to block the email. The Barracuda Spam Firewall also uses second-generation OCR engines that incorporate both animated GIF analysis and fuzzy logic to thwart spammers' attempts to obfuscate text inside images.

Further, the Barracuda Spam Firewall's multi-pass OCR engine performs deep analysis by performing image pre-processing operations to normalize the images prior to final optical character recognition passes. This additional processing is particularly effective against attempts to hide text by varying color and contrast combinations of text and backgrounds.

Automated Fingerprint Generation:

Barracuda Central collects image spam samples from its honeypot systems and from more than 40,000 customer systems throughout the world. Once collected, Barracuda Central processes the identified spam images through the multi-pass OCR engine, generates fingerprints for these images and packages them for automatic delivery to Barracuda Spam Firewalls through Energize Updates. By comparing fingerprints of images in subsequent emails against a database of known spam images, the Barracuda Spam Firewall can efficiently and intelligently block image spam analyzed at Barracuda Central.

By combining the tremendous image processing power at Barracuda Central with the efficient handling of fingerprints in the Fingerprint Analysis layer of the Barracuda Spam Firewall, customers benefit from latest image processing techniques while minimizing the processing requirements on their Barracuda Spam Firewalls.