xtract.bot

Image perceptual hash

Try it interactively →
POST /api/image-perceptual-hash

Compute a DCT-based pHash of an image. Robust perceptual fingerprint for de-duplication and reverse-image-search.

A DCT-based perceptual hash (pHash). Slightly more robust than `image-hash-blockhash` for non-trivial transformations like cropping or rotation, at the cost of being a bit slower to compute. Use the same way as `image-hash-blockhash` — compute the hash for two images and compare Hamming distance.

Inputs

NameTypeDefaultDescription
image*fileInput image (PNG only).

Response

Modes: json. Cache: yes (24h TTL).