A novel effective scheme for automated text detection and character recognition in natural scene images is presented in the paper. The proposed text detection approach belongs to the category of connected component-based methods utilizing Maximally Stable Extremal Regions (MSER) feature detector. Various literature based geometrical and contour oriented filters, used to distinguish between text and non-text MSER regions as well as to group remaining text regions into words and phrases, are applied first. Novel filters, designed to reject remaining non-text regions and words (phrases) that are not in line with assumed properties, are utilized next. Final words and phrases are recognized using an OCR system. Finally, an application of the presented approach within the IMCOP content discovery and delivery platform is briefly described.
CITATION STYLE
Baran, R., Partila, P., & Wilk, R. (2018). Automated text detection and character recognition in natural scenes based on local image features and contour processing techniques. In Advances in Intelligent Systems and Computing (Vol. 722, pp. 42–48). Springer Verlag. https://doi.org/10.1007/978-3-319-73888-8_8
Mendeley helps you to discover research relevant for your work.