Summary: | 碩士 === 義守大學 === 資訊管理學系 === 101 === In daily life will see all kinds of unfamiliar English words or word, often because we see these unfamiliar words and cause learning or communication difficulties, immediately translated into many people's needs, so this research to develop a " OCR System for text images capture by smart phone", this system is built on the smart phone, using phone camera functions with intuitive touch users to instantly find text in the image, so that people encounter in life will not words When the system can be used to find intuitive front of people do not see the position of words, and then translate to facilitate people to greater efficiency in the next instant access to knowledge, is a building belongs to smart phone, for English Instant fast text localization translation system.
This study English words often encountered blocks to build the system, simply divided into three parts, one, image pre-processing, the user wants to do image recognition image pre-processing, in order to facilitate the subsequent OCR, two, text block detection, the image pre-processing of images to do to complete the image text block detection part, and cut out to do identification, three, character recognition and translation, the text will be cut out using a class neural network to do the image of English text recognition and identification of the results to go with Internet translation services do more than right, will translate the results to try to understand the word meant, and learn.
In this thesis, the day will often encounter an English word to develop text localization translation system, the results can meet the user wants to locate and translate the success rate of 92%, enough to prove the correctness of systems development research.
|