A visualization method of huge video/image databases based on formal concept analysis and the attribute partial order theory

In order to improve the visualization of the whole relational structure for the image/video databases, a method based on formal concept analysis and the attribute partial order structure diagram is proposed. By choosing two groups of different attributes for the same object group, two attribute partial order structure diagrams are generated and connected. Then a normal and reverse serial model is obtained. And using huge video/image databases randomly selected from internet as the example, we show that the proposed method is effective to grasp the relational structure and to discover new hierarchical relationships.