提出了一种基于提升小波变换的快速多聚焦图像融合方法。首先利用提升小波算法将原始图像分解为四个子带: LL, LH, HL, HH, 后将代表三个方向高频细节子带LH, HL, HH采用提升小波反变换, 以获得各方向子带的高频细节图像, 采用高斯核权重算法计算所得到的高频细节图像的非均匀加权区域能量, 再根据基于能量的图像融合规则得到最终融合图像。对比了几种多聚焦图像融合方案的性能, 实验结果表明, 在融合效果相当的情况下, 文中方法比现有方法在处理速度上有明显的优势。

A fast multi-focus image fusion algorithm based on lifting wavelet transform is introduced. Firstly, the original images are decomposed into four sub-bands such as LL, LH, HL and HH through the algorithm. And then, high frequency details sub-bands such as LH, HL and HH representing three directions are performed lifting wavelet inverse transform to obtain high frequency details images of sub-bands on all directions. Non-uniform weighted area energy of obtained high frequency detail images is calculated through Gaussian kernel weight algorithm. And the final fusion images are obtained according to image fusion rule based on energy. Comparing with the characteristics of multi-focus image fusion schemes, experimental results show that the method is more effective than present methods on processing speed at the condition of same fusion effect.