There's a Microsoft product called PhotoDNA that does a fairly decent job of finding near matches. It's specifically used by several major image hosting places

There's a Microsoft product called PhotoDNA that does a fairly decent job of finding near matches. It's specifically used by several major image hosting places (Facebook and Tumblr I can specifically attest to using it, at least in the past) to find CP content uploaded by users. The hashes are shared among the services using PhotoDNA via NCMEC.