An Attention-Based Image Retrieval System

In this study, a new Content Based Image Retrieval system, called Attention-Based Image Retrieval (ABIR) is proposed by using the Itti-Koch visual attention model. For this purpose, first the regions are extracted by using the saliency maps. Then, the attention values, obtained from the saliency maps are used to define a new similarity metric. Finally, this metric is used to retrieve the similar regions. The proposed approach extracts the region of interests without employing any segmentation method. The extra emphasis given to high attention values, provide a better matching criterion, compared to the classical Euclidean distance. The power of the proposed method is demonstrated by the experiments. It is observed that ABIR retrieves with higher precision performance compared to the state of the art object based retrieval methods, in images with cluttered background.