Aiming at the problem of how to accurately detect the focusing region and how to overcome the registration error and noise sensitivity in the process of multi-focus image fusion, a multi-focus image fusion algorithm based on filter operator and double-scale decomposition is proposed. First, the algorithm performs Gaussian-Laplace filtering on the source image, and performs difference operation between the filtered image and the source image to separate out the high frequency information of the multi-source focused image. Then, an initial decision graph with complementary edge information is generated after decomposing the multi-source image edge and local high frequency information by the structure-based double scale focus measurement method. Finally, the initial decision graph is refined step by step based on the consistency test method to generate the fused decision graph, and the fused image is obtained according to the per-pixel weighted average rule. The experimental results show that, compared with other focusing strategies, this focusing region detection method has higher robustness and better recognition ability for different noises, and the processing time is less than 0.5 s.