The Research Of Color-based Image Retrieval

With the development of multimedia technology and Internet, more and more image resources are available and the need for image retrieval becomes more and more urgent. Due to the subjectivity and complicated of image labeled with text, the result of traditional text-based image retrieval technique is not satisfactory, thus giving birth to content-based image retrieval technique. The new technique acquires image features through analyzing the contents of images, and then finds the results by feature matching. Color is the most prominent image features, but also the main characteristics human eye perception to identify image with rotation invariance and scale invariance, so color-based image retrieval has been an active research area in image retrieval. The main contributions of this thesis are as follows:In this thesis, the fundamental theory of the content-based image retrieval and the situation of the research application are introduced. The content-based image retrieval techniques especially the color retrieval techniques and its current status are reviewed, along with some discussions of the key techniques of image retrieval.This paper presents a new image retrieval method based on color histogram entropy and main color of image blocks. In order to retrieve the image of the same color background but different color objects can take advantage of the color images features, information entropy will be introduced into image retrieval, image color histogram information entropy is an important feature of image similarity matching. At the same time, as the symmetry of entropy brought about some problems and cause some misuse of retrieval, so introduce the main color block approach. In order to highlight the middle of main image and limit the edge of background, use the thought of overlapping blocks, the joint distance of image entropy and the main color of image blocks as a standard of the similarity matching. The experimental results show that the precision rate and recall rate of this method is higher than the general methods in overall trend, especially the retrieval rate of the images with same color background but different color objects is improved the more.Designed and implemented a content-based image retrieval system, its function include image input, database management and image retrieval. It apply the theory of image retrieval in to the practical, as experimental platform for testing image retrieval algorithms. This system verifies feasibility and effectiveness of color-based retrieval algorithms proposed in this paper.