A Chinese Words Detection Method in Camera Based Images

Text in camera based image carries important information for visual content understanding and retrieval. Locating text from complex background is a challenge task and complex structure of Chinese words make it harder. In this paper, we develop an application which is targeted toward smart phone with camera. Our goal is to enable the smart phone to detect Chinese words in nature scenes. First of all, we implemente Stroke Width Transform [1] (SWT for short) of images on smart phones. Consequently, we improve the filtering process of detected connected components. Finally, we propose grouping methods of structure analysis to detect Chinese word in nature scenes. We have implemented the algorithm and tested it. Experiments show that our Chinese word detection approach works well.