We are developing a novel concept of real-time wearable translating glasses, named intelligent glasses, which can realize real-time multilingual translation. It is useful to a tourist who does not know the local language.
Let's imagine that a foreigner is using wearable translation robot during his foreign trip. The system can tell the wearer which building is a hotel, a restaurant, a bank, or a supermarket, which bus line should be selected, how to go to his destination according to the translation of road signs, which course is suitable for his taste based on the translation of menu, etc. All of these can make his journey much more delightful.
The system consists of a small head-mounted camera, a wearable computer, and a head-mounted display. These three components collect, translate, and present information about the texts. By introducing different Optical Character Recognition (OCR) systems and translation systems, we can realize the translation between different languages by means of flexible configuration.
In summary, the main contribution of this work is to create a novel wearable device using digital camera and wearable computer technologies for the enhancement of vision. This pair of intelligent glasses is an intelligent machine with features below:
(1) Translation: the glasses can translate the text in the camera’s field of view to different languages;
(2) Real-time: the process of text extraction, recognition, translation and displaying cost three seconds in general, and hence can perform in real-time;
(3) Wearable: this machine is light and easy to use and thus highly wearable. In this project, the most important and difficult problem is to detect and extract the texts from images.
We propose a new text detection method called character intrinsic characteristic-based (CIC-based) text detection algorithm. It performs well even in a complex environment. For example, it can extract a correct text string even though the digital camera gets a very vague image.