Overview
Fortium’s latest development for the detection of Image SPAM is Similis. Our image fingerprinting and matching systems are able to match images to a learning database. Using a proprietary fingerprint technique Similis accurately identifies images typically used in SPAM which have similar content and characteristics regardless of file format.
The integration of Similis into security applications provides development partners with ‘added value’ systems, offering end users protection against image SPAM and unwanted content. The advantage of the Similis technology over existing solutions in the market place is that the fingerprint technique provides much wider matching capabilities. We also provide high speed video frame, file retrieval and analysis scanning tools to search and retrieve the content.
Even when an image has been resized, cropped, watermarked, edited, reflected and rotated, Similis can detect images which match those in a defined database.
The Similis database is much smaller than traditional data-tracking systems as the fuzzy matching technique requires less data to match against thus saving time and resource. Similis technology is fast, accurate and easy to integrate.
Image SPAM
Current examples of image spam include blocks of text placed in an email as an image, with subtle changes to image background. This technique is defying not only the text engines but also standard fingerprinting solutions. SPAM has evolved in an attempt to fool standard SPAM OCR-based filters. The number of unsolicited messages containing images has grown significantly throughout 2006 and now represents 25 to 45 percent of all junk e-mail (Spam Doubles, Finding New Ways to Deliver Itself, NYT, December 6th, 2006).
Individual Approach
Ever concious to the increase in threats with regards to digital imagery, Fortium introduce Similis to successfully tackle the problem of bandwidth-hogging emails. Similis’ technique is able to successfully match these images which look identical but are not. Similis has shown in testing it can still images which vary up to 50% from the original. Complimentary to existing systems, Similis is the answer to the missing percentage of spam which meet customer expectations.
Image SPAM Example

The examples shown above illustrate how SPAM has evolved in an attempt to fool standard SPAM OCR and fingerprint matching filters. The image content is the same but with subtle differences in background colour and flecks.