Modeling visual attention mechanisms has been a very active area of research in past years owing to the challenges it poses. Many models exist, which have been successfully implemented in content based image retrieval systems. Owing to the vast quantities of image data available digitally, services for indexing or retrieving images based on queries have been gaining popularity. In this paper, a novel method is presented for image saliency detection using a more efficient color space model (performance-wise) based on the color distribution of the images instead of the primary visual features. It is done by combining global and local feature extraction into a single method of content detection within an image for purposes of image retrieval, which is proven to be more efficient, as well as taxonomy of various distance metrics used to identify local features. Furthermore, we gauge the performance of these metrics on a 9908 set of test images, based on their precision and recall. The paper has inferred the result by using a set of test images and evaluation methods that can serve to evaluate future metrics.