A Novel Approach for Automatic Detection and Recognition of Text from Tamil Signages

The objective of this paper is to design an efficient system for automatic detection and recognition of text from signages. It is especially designed for the people those are illiterate in Tamil speaking countries. This work mainly deals with major step of detection of text in signage’s from the input image using fast hybrid grayscale reconstruction. The second step deals with a technique for extracting features from the input image in order to get the rectangular board having text using toggle mapping morphological segmentation. The third step involves segmentation process that deals with segmenting the individual characters using row and column wise segments. And then our work moves with the Hidden markov model which plays an important role of extract information about the text. The viterbi algorithm is used for recognition because it yields a better result. By which our proposed method is expected to provide higher recognition accuracy than the conventional method.