论文部分内容阅读
Local visual attention mechanism (LVAM) is firstly presented and combined with bottom-up visual attention mechanism (VAM) to form a unified visual attention framework, based on which a novel infrared object detection method is proposed. Intensity feature is applied to build visual saliency map. Regional Shannon entropy density is used to determine the position of visual attention focus. Accurate object area is extracted with region growing technique imitating local VAM. Experiments demonstrate the outperformance of our proposed visual attention framework and infrared object detection method.