Trademark Image Retrieval Based On Multi-feature Extraction

Content based image retrieval is to perform the similarity retrieval according to the image features representing the image content, which may be extracted in the generic or specific domain. Specific domain based image retrieval can benefit from the useful domain knowledge and the corresponding retrieval results can meet the user's requirement better than the generic one. With the development of the market economy during these years, the application of the trademark proliferates, which brings difficulty to the checkup of the trademark. It needs to set up an accurate and efficient trademark image auto-retrieval system. The content based image retrieval technique has been widely used in the trademark domain. This paper presents the research mainly on content based trademark image retrieval, a method of trademark image retrieval based on Multi-Feature Extraction, and an implementation of trademark image retrieval system based on Multi-Feature Extraction.The sub-block based trademark image retrieval, which is popular at present, uses global and local image features to retrieve trademark image after it is divided into some blocks in the square or polar coordinate by whole image feature. But each sub-block image can't represent real local feature of original image, so its performance would be affected. In view of existing situation this paper proposes a method of trademark image retrieval based on Multi-Feature Extraction and Multi-to-Multi Feature matching. The method of sub-image segmentation in this paper is based on visual image features, it could make sub-image features give full scope. The main idea of this paper: Firstly, the trademark image is pretreated and transformed into binary trademark image. Then, it picks up the geometric image with clearly boundary of the trademark images according to connectivity as sub-image, extracts the shape features of each sub-image. Finally, the sub-image features which regard as independence features and the features from feature database accomplish Multi-to-Multi Feature matching. And then the trademark image retrieval is carried out.This paper presents a design of Trademark Image Retrieval System based on Multi-Feature Extraction which is aimed to be an experimental system. The system platform for supporting this experimental system is Windows XP Professional, and the developing environment is Visual C++ 6.0 and Microsoft Office Access 2003. Experimental results indicate that the proposed algorithm has improved efficiency and retrieval accuracy in such a great measure, and the retrieved results match human visual perception very well.