Phishing Detection via Image Hashing and Segmentation

The patent segments the website detection process into multiple independent hash computations, one for each image on the website. Instead of analyzing the entire website as a single unit, the system divides it into discrete image elements, computes individual hash values for each, and then aggregates these hashes. This segmentation approach maintains high detection reliability while reducing computational complexity by enabling parallel processing and incremental hash updates when images change.

The patent applies preliminary action by pre-computing and storing hash values for images from known legitimate websites before phishing detection is needed. When a website is evaluated for phishing, the system retrieves these pre-computed hashes and compares them against the hashes of images on the evaluated website. This eliminates the need to re-hash images during the actual detection process, significantly reducing processing time while maintaining high identification precision through accurate hash matching.

The patent implements partial action by comparing only specific image hashes rather than requiring complete website hash matching. The system identifies phishing websites by finding sufficient matches between image hashes on the evaluated website and those in the database of known websites. This partial matching approach simplifies operation while maintaining broad detection coverage, as it can identify phishing sites even when not all images are identical or when images have been slightly modified.