Hive UDF for phash

License

• "Anti 996" License

• ["Anti 995" License]

• ["Follow 955" License]

• ["World Peace" License]

• So, I'm waiting for the three licenses above to republic.

Backgroud

I used the ImageHash (GITHUB) in Python code. For easy to use, created the phash UDF for Hive for Java. The Java code refference from "Chinese Blog about Phash in Java"

Note:

Using median instead of avarage function.

Maybe some errors in the phash algorithm, please check it out when you actrually use it.

I'm unfamiliar with JAVA, by the way.

Speed

0.5s/image, so, it's UNusable in produtive envirment abviously. Luck!

Maven

PATH_TO_MAVEN/mvn clean package

then add the .jar to your Hive envirnment, and create the function.