Traditional picture dictionary provides only brief graphics for explaining the meaning of a word or the situation of its usage. To provide additional pictures with various contexts, movie screenshots are used in this illustrated English-Chinese lexicon system and methods of extracting these screenshots are developed. First the subtitle text is searched for the inquiry word to locate the corresponding movie scenes; then a local daemon process grabs the screenshot via DirectShow function calls. The grabber then returns the image to the web front-end using socket communications. Together with living dialogues, such kind of pictures is closely connected to the background movie stories and thus can provide rich context information for the dictionary user. This system has been deployed online as part of the Movie-English learning system.
Keywords: illustrated lexicon, movie, dictionary, DirectShow, software architecture.