提出了一种基于视觉熵的模拟生物视觉信息处理的视觉注意计算模型，在使用图像的低层特征的基础上，充分考虑了视觉熵中人眼对图像信息的主观量度。该模型首先将图像分成4×4的图像块，分别计算每个图像块的均值、方差、幅度和亮度变化率；采用中央邻域差算子得到各图像的特征图，合并成一个显著图；计算该图中每个显著区域的视觉熵，将最显著区域作为视觉注意的焦点，利用人类视觉注意的抑制返回机制，实现视觉注意焦点的转移，从而完成模拟人类视觉注意的整个过程。实验结果表明，该模型基本上能够模拟视觉注意的过程，而且复杂度较低，具有较高的实用价值。

A computational model of visual attention is proposed to simulate biologic visual information processing based on visual entropy. It is achieved by using low-level features of images and considering the human eye on the subjective measure of image information in visual entropy. First image is divided into 4×4 block images, images are calculated for each block of the mean, variance, range and rate of change in brightness. A center-surround operator is used to produce the conspicuous map of the images and incorporate them into a saliency map. Calculating each of the conspicuous regional’s visual entropy and regarding the most salient region as the focus of attention, the inhibition of return mechanism is used to transfer the visual focus to simulate the whole process of visual attention. The experimental results show that the model corresponds with the procedure of visual attention. Besides, it has low complexity and high utility value.