Image Retrieval Based on Sub Regions and Primary Colors
Yijun Hu · Journal of Hunan Industry Polytechnic · 2005
It is easy to perform for traditional image retrieval based on color histogram, however, its calculation in color histogram intersection is too large and the space distribution information is lost. In this paper, we propose a new method for image retrieval which is based on Sub Regions and Primary Colors. The image is divided into 9 equal sub-region, and the weight value of every sub region can be changed by users according to its importance. Finally the method for calculating the similarity is provided. This method not only simplifies the calculation in histogram intersection but also remains the space distribution information in color histogram to some extent. It is proved that this method is more effective to distinguish the object from the background and improves both the retrieval speed and the retrieval precision.