Ethiopic and Latin multilingual text detection from images using hybrid techniques

Abstract

Caption and scene texts found in images contain valuable information. These texts can be used for many applications to answer questions like what, when, where, and by who to give context to the images. So, automatic text detection enhances the user’s understanding of the media content. In Ethiopia, most street posts and promotional boards are written in multilingual characters
such as Latin (English, Afaan Oromo etc.) and Ethiopic (Amharic, Tigrigna etc.). In this work, we have studied Ethiopic and Latin
multilingual text detection from images for both caption and scene texts. After the images are pre-processed, maximally stable extremal region (MSER) algorithm, aspect ratio and stroke width transform (SWT) algorithm are used to extract text regions, respectively. Then texture features are computed using local binary patterns (LBP) from the extracted regions. Finally, the support vector machine (SVM) is used to classify text region vs nontext using the computed LBP features. We prepared a new multilingual Ethiopic and Latin script image dataset to evaluate our method.