：根据小波低频子带图像的轮廓模糊度和高频子带图像的细节信息量随小波分解层数增多而增多的规律，从图像熵

出发，提出了一种基于低频子带图像熵差的最佳小波分解层数选择法, 该方法通过计算不同分解层数下各低频子带的图像

熵差，选择最接近原始图像熵差的分解层数作为最佳分解层数。用多种基于小波变换的图像融合法分别对两组医学图像进

行仿真实验，结果表明根据该分解层数选择法得到的融合图像目视效果最好，与相关系数、峰值信噪比、模糊Chebyshev 距离

值等客观评价指标保持了很好的一致性。

-Abstract：According to the law that the contour fuzziness of the low-frequency sub-band and the detail information of the high-frequency

sub-band increase with the growth of the wavelet decomposition levels, an effect selection method of the optimal decomposition level is

proposed. Firstly, the difference of original image entropy is calculated. Then, the differences of the low-frequency sub-image entropy

based on different decomposition level are calculated. Finally, the decomposition level which makes the difference of the low-frequency

sub-image entropy closest to the difference of original image entropy is found. This level is the optimal decomposition level. To fuse

two groups of medical images, simulation experiments are done by use of the wavelet transform and the PCA transform based on wavelet.

The experimental result indicates that the fusion image used by the proposed method has the best subjective feeling of human being, and

it has very good compatibility to objective