Precision Improvement of Content-based Image Retrieval Using Dominant Color Histogram Descriptor

Searching for similar images in an image database color is the most often considered feature. The color-based comparison and retrieval has a variety of widely used techniques. In this article a new color feature is proposed to use, that takes into account the dominant colors in HSV color space, the color histogram and the spatial distribution of pixels. With the proposed feature experiments were carried out in two image databases, and can be stated that the precision of retrieval has significantly improved. During the experiments the applied parameters were compared, and the most suitable parameter values are determined. Key–Words: Content-based image retrieval, Color features, Color histogram, Dominant color histogram, Parameter determination