A hybrid approach to extract scene text from videos | IEEE Conference Publication | IEEE Xplore