Text Detection in Natural Images Using Bio-inspired Models
Konstantinos Zagoris, Ioannis Pratikakis · 2013
Textual information in images constitutes a very rich source of high-level semantics for multimedia indexing and retrieval. In this paper, a new approach is proposed for detecting text in natural images inspired by the properties of the Human Visual System (HVS). In particular, the detection is based upon the functionality of the OFF and ON center-surround cells of the HVS. Cells are combined at different scales to guide a both efficient and effective algorithm. Performance evaluation relies on the ICDAR 2011 Robust Reading Dataset.