Study of embedded font context and kernel space methods for improved videotext recognition | IEEE Conference Publication | IEEE Xplore