Application of Rough Set Theory on scene image classification

This paper utilizes the rough set theory to classify the scene image. The system learns knowledge for classification automatically and therefore breaks the limitation of the traditional template method. For a scene image, its color relates with object and semantic closely. Therefore, we extracted two major colors and the quantity, spatial relations and textures of the regions formed by them to describe the scene image. Experimental results show that the rules are effective to classify the four types of scene images and obtain 85% of average retrieval performance.