A Modified Histogram Approach to Trademark Image Retrieval

Trademarks are considered as valuable intellectual properties. They play very important roles for successful business or companies. The logos or trademarks are also very significant objects in consumer world applications, because they are specially designed marks to identify and represent not only the quality of actual products but also the reputation of companies, products and services. With the continuously increasing number of registered trademarks, how to avoid designing a new trademark similar to an existing the registered trademarks or logos becomes an important problem. To treat this problem developing an automatic and efficient content based trademark retrieval system is imperative. In this paper, we propose a modified composite histogram approach for trademark image ranking and retrieval. The histogram represents the composition of distance and angle pair-wise histogram as a feature vector of trademark shape for retrieving similar registered trademarks. Then, based on the rank of the feature distance, a similarity measure is provided to do the similar trademark retrieval. Experiments have been conducted on registered trademark databases. Impressive results are shown to demonstrate the robustness of the proposed approach. Moreover, it is quite simple to construct the distance–angle pair-wise histogram for a trademark object.