Research On Content Based Trademark Image Retrieval

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 market economy the trademark plays the more and more important role in the current society and the content based image retrieval technique has been widely used in the trademark domain. This paper does research mainly on content based trademark image retrieval, presenting three new trademark image retrieval methods and a human-computer interactive trademark image retrieval system based on multi-feature fusion.Firstly, this paper presents the trademark image retrieval method based on distance distribution histogram. The image object region is extracted according to the object pixels' circum-circle and then partitioned based on the concentric circles to produce a series of sub-images, according to which the distance distribution histogram is produced and then normalized. Secondly, this paper presents the trademark image retrieval method based on region orientation Fourier transform. The edge Fourier transform is an important shape descriptor, but it makes use of only the image edge information ignoring the image region information. To overcome the disadvantage of the edge Fourier transform, this paper presents the region orientation Fourier transform and applies it to the trademark image retrieval. Lastly, this paper presents the trademark image retrieval method based on sub-block image features. The image is partitioned into several sub-blocks, each sub-block image features reflect the image local information and the combination of multiple sub-blocks' image features describe the whole region shape. The hierarchical from-coarse-to-fine sub-block-partitioning structure can describe the image shape feature from the multiple levels. The comparison experiments show that the trademark image retrieval method based on distance distribution histogram is superior to the one based on single Hu invariant moment, the one based on region orientation Fourier transform is superior to the one based on Hu invariant and the one based on edge Fourier transform, and the one based on sub-block image features is superior to the one based on Hu invariant and the one based on grid image features.Currently there is no perfect feature for content based image retrieval to make the retrieval result meet all the people's visual perception and all the applications' practical requirements very well. So it is necessary to fuse the multiple features and perform the human-computer interaction in the image retrieval. The three trademark image retrieval methods presented in this paper are almost the same with each other under the evaluation of visual consistency, but the corresponding retrieval results have different styles. All of them are absolutely necessary as the effective alternatives provided to the user in the retrieval system based on multi-feature fusion. This paper designs a human-computer interactive trademark image retrieval system based on multi-feature fusion, which is experimental system and can be used to test different retrieval algorithms or fuse them to perform the image retrieval.