中文摘要：

从人眼的视觉机制出发,提出了一种拟人视觉系统的显著性检测方法。该方法首先对图像进行量化并选取出高频颜色,降低了计算的复杂度,然后对图像进行分割对比,初始的视觉点以图像的中心作为基准点,通过提出的视觉引力模型迭代计算出视觉点的移动轨迹,最终寻找到显著区域。在公开的数据集上的实验结果表明,本方法所寻找到的显著区域相对于其他方法能够更精确地对显著区域进行标注,更加符合实际应用。

英文摘要：

This paper presents a saliency detection based on human visual systems. To begin with, it quantifies image and chooses high frequency color to reduce the complex of calculation. Then, it splits image and chooses the center point of image as the first visual point, iteratively calculates visual point according to visual attraction model, and acquires saliency area. The experimen results show that the method is more accurate than other methods to mark the salient and is very useful for practical application.