Background Removal in Image Indexing and Retrieval

This paper presents our research in image content based indexing and retrieval, a key technology in digital image libraries. In most of the existing image content-based techniques, image features used for indexing and retrieval are global, features are computed over the entire image. The major problem with the global image feature based retrieval methods is that background features can be easily mistakenly taken as object features. When a user attempts to retrieve images using color features, he/she usually means the color feature of an object or objects of interests contained in the image. The approach we describe in this paper utilizes color clusters for image background analysis. Once the background regions are identified, they are removed from the image indexing procedure; therefore, no longer interfering with the meaningful image content during the retrieval process. The algorithm consists of three major steps of computation, fuzzy clustering, color image segmentation, and background analysis.