Extracting Bottom-Up Attention Information Based on Local Complexity and Early Visual Features

During the last few years, extraordinary progress has been made in understanding the basic principle of how information is processed by visual cortex. This brings more and more concerning to bottom-up attention at home and abroad, and a large number of models based on it are built. But difficulties are met about how to extract efficient bottom-up attention information which is invariant to image scale, rotation and translation. Inspired by the research of visual attention in psychology, a novel algorithm for extracting bottom-up attention information （integration of local complexity and early visual features，LOCEV） is proposed in this paper. Bottom-up attention information is composed by saliency of certain regions correspond to each point in image, and scale of the region varies with complexity of local features adaptively. New saliency metric is defined as a product of three terms: local complexity, statistical dissimilarity and early visual features. Saliency of certain regions corresponding to all points in image is defined as bottom-up attention information. Salient regions are salient both in feature space and over scale. The extracted bottom-up attention information is invariant to image scale, rotation and translation, and is shown to be robust to noise. An attention model is developed based on this algorithm. Experiments results with natural images demonstrate its effectiveness.