Color Image Compression Using Data Clustering Techniques

The importance of Image processing has increased greatly with the development of the World Wide Web. Image Compression techniques allow an image file to be decreased in size while retaining the original quality of the image, thus facilitating quicker and more efficient web browsing. Clustering technique can be used to find the best palette for representing the original colors in the image. Two methods of clustering are explored in this paper. The k-means algorithm and the winner-take-all algorithm, both use an original set of cluster centers to form groups and find new centers. These two algorithms require a great amount of computation, which generally decreases as the specified number of cluster centers decreases. Thus, another tradeoff for preserving image quality is to incur more computational time in transforming an image.