An Efficient Two-Stage Trademark Retrieval System

Abstract

Recently, content-based trademark retrieval has received wide attentions. Contour and region are two important attributes in similarity measure of two trademarks. This paper presents a two-stage trademark retrieving system that integrates contour and region attributes. In the first-stage, the contour signature of an input trademark is extracted and used to filter out unlikely matched candidates from the database. In the second stage, the region feature represented by MPEG-7 ART is used to search the best match from the possible candidate trademarks. To further improve the retrieval performance, a simple user interaction scheme is added in the second-stage matching. The simulation results indicate that the proposed method improves retrieval accuracy and speed significantly, as compared to ART.