Text/graphic separation using a sparse representation with multi-learned dictionaries | IEEE Conference Publication | IEEE Xplore