Scene text recognition using touch screen interface

Abstract

Text separation in natural scenes is a crucial step to recognize scene text. Since computational power in a mobile device is limited, current text extraction methods are impractical in real-time devices. We propose an efficient text extraction method and scene text recognition system by utilizing user`s indication. When the user simply draws the line on touch screen, the system can extract text in natural scenes efficiently using this information. Text region is estimated from user`s touch and color candidates are extracted. After that, the system chooses color candidates which have high probability of texts. By using the assumption that the text components are aligned along the line, we verify the true text region between text candidates. Finally, we expand text region to deal with partially drawn lines and severe color gradation. In the experiments on ICDAR 2003 database, our method demonstrated effective performance. Additionally, we showed usability of our method in a portable device by implementing demonstration system.