| With the rapid development of society and the continuous improvement of information, a large capacity image database appears. How to use image features to retrieve image effectively is a hot topic in the research of the content-based image retrieval.Considering that the MPEG-7 dominant color descriptor ignores the interference of background noise on the target image, an improved dominant color descriptor (IDCD) is proposed. By retrieving the boundary area of the image, the improved method extracts the background noise to reduce the background noise on the target image. Experimental results demonstrate that the IDCD can achieve better results on the retrieval accuracy and retrieval efficiency.The single feature of image can not reflect the image information, so the retrieval method by using the multi-features is proposed based on center block. The traditional color feature has been improved, which effectively distinguishes between the target object and the background noise in the image to reduce the interference of background noise in calculating the image distance. Texture feature based on central block, avoiding the background texture in the image, is closer to the texture feature of the target object. We integrate the two features of the image in center block by external Gaussian normalization. With the same kinds’ integrated feature retrieval, the retrieval accuracy has greatly improved by the analysis of a large number of experimental results.The experimental results show that integrated multi-characteristics of image can improve the retrieval precision of similar images. But while integrating multi-features to retrieve images, weight setting is a problem. In order to solve the fixed weight problem of image retrieval we use the differential evolution algorithm to optimize weights in the thesis. The experimental results demonstrate that retrieval algorithm with optimal weights has a higher accuracy than fixed weights on image retrieval. |
PDF Full Text Request