Automatic caption localization in videos using salient points | IEEE Conference Publication | IEEE Xplore