Datong Chen
In this paper, we present a method to improve video OCR with multiple character hypotheses. The text regions in video need to be binarized before work as the input of current OCR system. Tranditional binarization do not use any structural information about ...
IDIAP2000