Abstract: For recent years, many multi-focus image fusion methods have the same problem that a good image fusion is always accompanied by a great quantity of computing and complex process. Inspired by the traditional pixel-level sub-block fusion method and the theory of differential approximation in mathematics, a multi-focus image fusion method based on pixel-level sub-block successive approximation is proposed. Experimental results have showed that: on the fusion effect this method has reached or even been slightly better than some of the methods proposed in some literature, but on the computation and the program run time this method is much better than other image fusion methods and this method is more conducive to rapid implementation of the multi-focus image fusion in the practical application.

Key words: multi-focus image, image fusion, sub-block successive approximation, spatial frequency

摘要： 针对近年在对多聚焦图像融合领域中存在图像融合效果好的方法运算量大、实现起来复杂的问题，在传统的像素级分块融合方法和数学微分中逼近原理的启发下，提出了多聚焦图像在像素级分块逼近的图像融合方法。实验结果表明：该方法在融合效果上已经到达甚至略优于一些文献中提出的方法，而在运算量和运算时间方面要明显优于这些方法，更有利于实际应用中多聚焦图像融合的快速实现。

关键词: 多聚焦图像, 图像融合, 分块逼近, 空间频率