中文摘要：

在分析和比较拉普拉斯金字塔与低通比率金字塔图像融合方法的基础上，结合对于人类视觉特性的了解．提出一种基于视觉阈值特性的金字塔图像融合方法．该方法要求当目标亮度位于其阈值之上并且不同源图像目标对比度极性相同时，采用对比度指标确定融合要素；否则，采用绝对值指标确定融合要素．实验结果表明，该方法既可以增强融合图像的对比度信息，又能有效避免低通比率金字塔方法中有时出现的斑点噪声，所需的额外计算量小，因此适用于要求实时性的融合领域．

英文摘要：

A visual theshold based pyramid image fusion method is proposed after the Laplacian pyramid method and rate-of-low-pass pyramid method were analyzed and compared. It adopts the following fusion strategy： when the targets from different source images are above their individual thresholds and their contrast polarities are found to be identical, the corresponding fused elements are determined according to contrast information; otherwise, the fused elements are determined according to absolute difference values. The results showed that the contrast information of fused images can be enhanced and the speckle noise frequently incurred in the rate-of-low-pass pyramid method can be avoided at the same time. Its extra computational load is not significant, so the method is suitable for use in some real-time fusion fields.