A STUDY OF IMAGE RETRIEVAL BASED ON THE K-MEANS CLUSTERING TECHNIQUE

In this paper, we evaluate the effectiveness of a clustering technique for the fast retrieval of faces in a large image database. Our approach is based on partitioning of the images in the database through k-means clustering. The success of the technique in decreasing the time needed to match an image of a wanted person to those in a database of images is investigated. Two steps are needed towards this aim. First, a fast distribution of images from a large database into groups is applied and evaluated using the k-means clustering technique. Second, an image retrieving method is used and tested to find a match to the image of the wanted person within the large database