To perform regular expression queries on non latin characters you need to reference their character types. In the case of the Chinese, Japanese, Korean characters sets, use the \p{IsCJKUnifiedIdeographs} selector.
For example, to capture only CJK characters from a mixed language text block, use a regex something like below...
REGEX(\p{IsCJKUnifiedIdeographs}+)
Here is a full list of language classes that are available.
Link to this article http://umango.com/KB?article=94