图像关注焦点(FOA)检测是基于人眼视觉关注模型的图像感兴趣区提取的关键技术。为了更加精确、合理地搜索图像关注焦点，提出一种基于双阈值视觉关注模型的FOA检测算法。算法首先提取图像的亮度、颜色、方向和离散矩变换(DMT)特征，生成各个特征对应的特征图；然后将各特征图合并为一张包含多种特征的显著图；最后根据显著图的灰度直方图建立静态阈值与动态阈值，确定图像关注焦点的位置与数量。实验结果表明：新算法在图像关注焦点检测的准确性上较Itti模型有更为优秀的表现，更符合人眼视觉的关注习惯。

Image focus of attention detection is the key technology of image region of interest extraction based on human visual attention model. In order to search the image focus of attention more accurately and rationally, a new image focus of attention detection algorithm is proposed based on double threshold visual attention model in this paper. First, the intensity, colors, orientations and the center moments of discrete moment transform are extracted to generate the conspicuity maps for each feature. Second, the conspicuity maps are combined to get a new saliency map. Finally, according to the gray-scale histogram of the saliency map, the static threshold and dynamic threshold are found to confirm the positions and numbers of the image focuses of attention. The experimental results show that the new algorithm has the more outstanding performance in accuracy of image focus of attention detection than the Itti's model, and coincided better with human's visual habits.项目(OFSLRSS201001)