Automatic text segmentation and text recognition for video indexing

Abstract

Efficient indexing and retrieval of digital videois an important function of video databases. One powerfulindex for retrieval is the text appearing in them. It enablescontent-based browsing. We present our new methods forautomatic segmentation of text in digital videos. The algorithms we propose make use of typical characteristics of textin videos in order to enable and enhance segmentation performance. The unique features of our approach are the tracking of characters and words over their complete duration ofoccurrence in a video and the integration of the multiplebitmaps of a character over time into a single bitmap. Theoutput of the text segmentation step is then directly passedto a standard OCR software package in order to translate thesegmented text into ASCII. Also, a straightforward indexingand retrieval scheme is introduced. It is used in the experiments to demonstrate that the proposed text segmentationalgorithms together with existing text recognition algorithmsare suitable for indexing and retrieval of relevant video sequences in and from a video database. Our experimentalresults are very encouraging and suggest that these algorithms can be used in video retrieval applications as well asto recognize higher level semantics in videos.