Abstract: The existing saliency detect methods do not consider the influence of human visual focus which affects the result seriously. This influence is especially obvious in medical image analysis. For example, the saliency of the same image is totally different in the view of different doctors because of the different kinds of diseases and different organs. To solve this problem, the spacial focus coefficient and gray focus coefficient are introduced and the saliency detect method based on visual focus of spacial and gray is presented. The experiment and the comparison with other method show that the results are different in different focus, which can help people concentrate on different salient regions when analyzing the medical images. The results prove the efficiency and flexibility of this method.

Key words: saliency, spacial focus, gray value focus, attention, medical image

摘要： 现有的显著性检测方法无法反映出因人类注意力焦点的不同而导致的显著性检测结果的不同，这些不同结果的影响在医学图像分析中尤为明显；例如，同一幅图像在不同医生眼中因检查不同疾病、不同组织器官，其所关注的焦点是不同的，进而“显著性”检测结果也应当有很大的差异。在计算灰度直方图对比度的基础上，引入空间焦点系数和灰度焦点系数，以反映人类视觉注意力在不同空间焦点和灰度焦点条件下对图像显著性检测的不同影响；提出了基于视觉空间注意力焦点与灰度注意力焦点的显著性分析模型。通过在不同医学图像上的实验，以及与典型显著性检测算法的比较分析，基于视觉空间注意力焦点与灰度注意力焦点的显著性分析模型在不同焦点下可以得到不同的分割结果，有助于将图像分析的重点聚焦在显著区域，实验证明了该模型的灵活有效。

关键词: 显著性, 空间焦点, 灰度焦点, 注意力, 医学图像