Introduction…<br /> Our project ‘Mobile camera based text detection and translation’ retrieves text from an images and converts it into text format, then it is translated to specified language.<br />3<br />
History<br /><ul><li> In 1929, first OCR device was invented but it was mechanical device
In about 1965, earliest form of OCR was implemented in one of the first generation computersfor Airline Ticket stock.
Revolutionary in 1971, it was implemented in postal services OCR systems where reading and printing of routing bar code was done on the postal code.
In 1974, the modifications was done which would allow blind people to have a computer read text to them out loud.
In late 90’s, Webcam was used for OCR process.</li></ul>4<br />
Present<br /><ul><li>Webcam integrated with computers are being used for capturing image and easily the text can be extracted by it and than translated.
The image can be analyzed and translated also online.
Only some software companies manufactures the OCR system in mobile , having high specifications.
ABBYY Mobile OCR, is the leading manufacture of mobile OCR.</li></ul>5<br />
Edge Detection<br />1. Convert Input image into Gray – scale image :<br />Y = 0.299R + 0.587G + 0.114B<br />2. Apply Blurring on image Y .<br />3. Find threshold value of Y2 =<br />9<br />
Text feature filtering:<br />10<br />After Detecting Text Area, the Extraction of the character from the image is perform<br />For Extraction & detection of the character the Edge detection, corner detection used.<br />
Requirement<br />Mobile Hardware Requirements:<br /><ul><li>ARM 11 processor or higher
Test & Results<br />Font :<br /> Recognition rate does not vary as font changes<br />Font size : <br /> As the size of text varies , Recognition rate will vary i.e. if the text is of larger size then recognition rate will be greater.<br />14<br />
Test & Results<br />Image quality :<br /> As image quality degraded recognition rate will decrease<br />Recognition rate of character ‘A’ , ‘B’ , ‘L’ will be higher than recognition rate of character ‘y’ , ‘u’ , ‘c’.<br />Fig. d: Test & result<br />15<br />
For translation of extracted text , Internet connection is required.
Translated text may have Grammatical mistakes</li></ul>20<br />
Conclusion<br /> This project which we have implemented is an Android Mobile OS based application which is web based real time mobile application for real-time text extraction, recognition and translation. <br />21<br />
Bibliography<br />Michael Hsueh “Interactive Text Recognition and Translation on a Mobile Device “ [Technical Report No. UCB/EECS-2011-57 ]<br />YassinM.Y.Hasan and LinaJ.Karam “Morphological Text Extraction from Images” IEEE Transaction on Image Processing Vol.9 No.11, Nov 2000<br />Nobuyuki Otsu, A threshold selection method from gray-level histograms. IEEE Trans.Sys.,Man., Cyber 9(1):62-66<br />Celine Mancas-Thillou, Bernard Gosselin, Color text extraction with selective metric based clustering. Computer Vision and Image Understanding 2007 <br />B. Epshtein, Detecting Text in Natural Scenes with Stroke Width Transform. Image Rochester NY, pp. 1-8.<br />Derek Ma , Qiuhau Lin, Tong Zhang “Mobile Camera Based Text Detection and Translation” – research paper<br />WWW.wikipedia.org/optical_character_recognization<br />22<br />