Research On Color Image Retrieval Based On Visual Attention Region

With the development of multimedia technology and the application of Internet, a great deal of digital images are produced by different fields everyday.How to obtain a desired image from the mass image database quickly and effectively has become a serious problem to be solved.In the case,the research on CBIR(Content-based image retrieval )was developed,and became a hot spot in the field of information retrieval. In this dissertation, the exploratory research work has been done around some key techniques of CBIR. The main contributions of this dissertation are summarized as follows:(1) In general,it is difficult to obtain ideal and stable retrieval result only using single feature in CBIR system,so we propose a new and effective image retrieval scheme based on multiple features in this paper. Firstly, the image is predetermined by using fast color quantization algorithm with clusters merging, and then a small number of dominant colors and their percentages can be obtained. Secondly, the spatial texture features are extracted using a steerable filter decomposition, which offers an efficient and flexible approximation of early processing in the human visual system. Thirdly, the pseudo Zernike moments of an image are used for shape descriptor, which have better features representation capabilities and are more robust to noise than other moment representations. Finally, the combination of the color, texture and shape features provide a robust feature set for image retrieval. Experimental results indicate that the proposed technique outperforms other schemes in terms of retrieval accuracy and stability.(2) The VAR(Visual Attention Region) detection plays an important role in bridging the gap between the image content and the high-level concept.In this paper,we design a VAR-based approach to CBIR. Firstly,An efficient algorithm to detect the VAR at object level is adopted. Afterwards, both features comprising color feature and texture feature are extracted from the VAR. The color distribution and image bitmap are used as color feature.Rotation-invariant and scale-invariant Gabor representations are used to descript the texture feature.The scheme only takes account of the features extracted from VAR instead of the global image. Experimental results show that the proposed image retrieval is more accurate and efficient in retrieving the user-interested images.