Research And System Implementation On Trademark Retrieval

When a new trademark is registered,it will be checked that it is same or very similar to the trademarks in the database ,that will determine whether it have qualification for registration,due to the increasing number of registered trademarks, trademarks retrieval based on artificial method shows some drawbacks,its workload is heavy and it is difficult to meet the real-time and accuracy requirements,so research on more accurate and more efficient trademark image retrieval methods and the establishment of automated trademark retrieval system in the field of trademarks is of great significance.The paper analyzes the classification retrieval method ,text-based retrieval method and content-based retrieval method, text-based trademarks and mixed trademarks are retrieved by text-based retrieval method, graphical trademarks are retrieved by content-based retrieval method, the paper mainly studied the trademark image retrieval method based on region and sub-block,it combines global features and local features,the blocked approach based on the circular area in paper is better than one based on rectangle area,because it isn't impacted by the changes in image rotation ,besides , this paper proposed a method based on the regional distribution of a trademark image, using the concentric circle splitting strategy based on the centroid and radius circle target region ,extracting its various sub-regions normalized entropys and their first-order difference scores as the shape feature, the experiments proved that the methods have a certain degree of retrieval performance and geometric invariance.Finally, a trademark retrieval system is established and made as an experimental platform , the various retrieval methods are authenticated.