Color is one of the most important low-level features used in image analysis and retrieval. In this paper we introduce an algorithm for extracting compact color feature in the CIELAB color space. Our method is based on an ant colony clustering algorithm, which helps us develop a perceptually based image similarity metric. We propose a dominant color and area percentage model, which provides a very compact color feature representation of images. Besides image similarity metric is developed based on dominant colors matching. The algorithm introduced in this paper is well suited for creating small color descriptors and is efficient. It is suitable for image representation, matching and retrieval. It also has some important advantages for content-based image retrieval system, such as great reduction of the feature file sizes needed for storage and indexing.