Word-Level Script Identification from Scene Images

摘要

Script identification on camera based bus sign boards are presented in this work. The text localization is achieved by a series of morphological operations. Number of texture features, such as gabor features, log-gabor features and wavelet features are extracted from the segmented text images to classify the text images into three scripts, English, Kannada and Malayalam. Three different classifiers are evaluated and results are reported in this work.