A hierarchical feature search method for wine label image recognition

Le résumé fourni par la source

In view of the people dramatic increase in the demand for wine, wine culture has be gradually integrated into Taiwanese's daily diet of life. However, the kernel skills of existed wine information systems are mostly based on text searching techniques and this will restrict the system usability especially the text on wine label is not print in English. There are only a small number of mobile information systems which provide related wine information with visual search function. In this paper, we proposed an efficient mobile visual search scheme for wine label recognition using a client-server architecture. Furthermore, a hierarchical feature search method is proposed to improve the image matching performance of searching similar wine label images in database. Experimental results reveal that the proposed wine label information system can achieve 95.05% accuracy rate in 202 wine label images recognition test and 0.26 second matching time for each image query.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.