Mobile Camera Based Text Recognition and Translation
Total Page:16
File Type:pdf, Size:1020Kb
Mobile camera based text recognition and translation EE 368 Final Project Proposal, Spring 2011 Derek Ma ([email protected]) Qiuhua Lin ([email protected]) Tong Zhang ([email protected]) Motivation This project is inspired by the well‐known iPhone app “Word Lens”. It identifies, recognizes the text captured by the camera on a mobile device, translates it to another language, and renders the translation onto the screen. The motivation of this project is to enable the user to navigate in a foreign language setting with the assistance of a mobile device. For instance, a Chinese tourist travelling in U.S. would be able to easily get instant translation from his Android phone app on public signs, restaurant menus, etc. Goal The goal is to achieve near real‐time mobile‐based system of text extraction, recognition and translation. We are to explore different text extraction algorithms for the OCR engine to achieve high recognition rate. The algorithms will be run on a Motorola Droid phone, with the OCR / translation engines run on a server. Approach We have considered two possible approaches for text extraction: (a). Identify text region automatically; (b) User assistive text region identification. After this step, following procedures will be applied: Return the binarized text image from text extraction algorithm‐> send to tesseract OCR engine on the serve ‐> Google translate application ‐> send translation back to mobile device ‐> display result in a user‐ friendly manner. Due to the limited resources, we initially assume that the text images to be translated are captured indoor with monochromatic background and enough light illumination, and the orientation of the text image should be constrained within a small angle of tilt. Constraints may be relaxed if time permits. Reference [1] Farshad Ghazizadeh, Optical Character Recognition, US Patent: 5,007,809. [2] Huiping Li, David Doermann, Omid Kia, Automatic Text Detection and Tracking in Digital Video, IEEE Transaction on Image Processing Vol. 9 No. 1, Jan 2000 [3] Yassin M. Y. Hasan and Lina J. Karam, Morphological Text Extraction from Images, IEEE Transaction on Image Processing Vol. 9 No. 11, Nov 2000 [4] Celine Mancas-Thillou, Bernard Gosselin, Color text extraction with selective metricbased clustering, Computer Vision and Image Understanding 2007 .