Text localization in natural scene images using multiple classifiers

摘要

Localization of text areas in images of natural scenes is essential for many applications, including support for the blind, tourist support, traffic control and content-sensitive image or video databases. Algorithms for localization of scene text must overcome issues of non-uniform scene lighting, scene text variations and distortions etc. In the paper we present the results obtained for scene text detection algorithms operating on various image representation domains, as well as our work on an algorithm based on an ensemble of simple classifiers.