An easy approach to recognize Tibetan script by geometric features
Wei Zhou, Liangyu Chen, Zhenbing Zeng · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2011
In this paper, we present a new approach to recognize Tibetan words by analyzing the structure and geometric features. The printed Tibetan words, including consonant and vowel symbols, are firstly generated by FontLister tool and through pre-processing, such as skeletonization and deburring. Then the words are splitted into the single characters, which can easily recognized by feature extraction and pattern matching. We adopt six features according to the geometry and stroke structure. Finally, the Tibetan words are automatically "understood" and converted into Latin alphabet transcription, which make more people read. The experiments show this approach has a recognization accuracy over 89%.