Text Localization in Real-World Images Using Efficiently Pruned Exhaustive Search

• Published in: 2011 International Conference on Document Analysis and Recognition pp. 687 - 691
• Main Authors: Neumann, L., Matas, J.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2011
• Subjects: Character recognition; Detectors; Image recognition; real-world images; Robustness; text localization; Text recognition; text-in-the-wild; Training; Vegetation
• ISBN: 1457713500; 9781457713507
• ISSN: 1520-5363
• DOI: 10.1109/ICDAR.2011.144

An efficient method for text localization and recognition in real-world images is proposed. Thanks to effective pruning, it is able to exhaustively search the space of all character sequences in real time (200ms on a 640 × 480 image). The method exploits higher-order properties of text such as word text lines. We demonstrate that the grouping stage plays a key role in the text localization performance and that a robust and precise grouping stage is able to compensate errors of the character detector. The method includes a novel selector of Maximally Stable Extremal Regions (MSER) which exploits region topology. Experimental validation shows that 95.7% characters in the ICDAR dataset are detected using the novel selector of MSERs with a low sensitivity threshold. The proposed method was evaluated on the standard ICDAR 2003 dataset where it achieved state-of-the-art results in both text localization and recognition.