Perceptual hashing helps catch reused ID photos in KYC fraud checks
Fraudsters often reuse one identity document across many accounts, altering the image slightly so each file differs. Cryptographic hashes like SHA-256 fail to flag these edited copies, since any byte change alters the entire hash. The article describes dHash, a perceptual hash that compares brightness gradients in a 9x8 greyscale version of an image, so near-identical pictures produce hashes differing by only a few bits. Testing showed resizing and JPEG recompression leave the hash unchanged, while cropping and rotation shift it, which is why documents should be normalised before hashing. Hashing only the portrait rather than the whole card avoids false matches between different people sharing the same card template.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.


Discussion (0)
Log in to join the discussion and vote.
Log in