Abstract: An improved multi-focus image fusion method based on wavelet transform is proposed. Multi-scale decomposition is performed on source images using wavelet transform to get high-frequency and low-frequency images. A method based on neighboring region variance weighted-average is applied to high-frequency image to get the high-frequency fusion coefficient；and a method based on local region gradient information is applied to low-frequency image to get the low-frequency fusion coefficient. The inverse wavelet transform is utilized to obtain fused image. The fused image by the proposed method is evaluated with some parameters such as root mean square error, entropy and peak signal noise rate, in comparison with traditional fusion methods. The experiment results show that the proposed method is effective on improving the effect and quality of fused image.

Key words: image fusion, wavelet transform, multi-focus image, neighboring region variance, gradient information

摘要： 提出了一种改进的基于小波变换的多聚焦图像融合方法。该方法采用小波变换对源图像进行多尺度分解，得到高频和低频图像；对高频分量采用基于邻域方差加权平均的方法得到高频融合系数，对低频分量采用基于局部区域梯度信息的方法得到低频融合系数；进行小波反变换得到融合图像。采用均方根误差、信息熵以及峰值信噪比等评价标准，将该方法与传统融合方法的融合效果进行了比较。实验结果表明，该方法所得融合图像的效果和质量均有明显提高。

关键词: 图像融合, 小波变换, 多聚焦图像, 邻域方差, 梯度信息