Abstract: A novel multi-focus image fusion algorithm is proposed for multi-focus image fusion. Pulse coupled neural network (PCNN) and rough set theory are employed to solve the problem of multi-focus image fusion. The clarity of original image is calculated and processed by PCNN. The original image pixels are classified based on rough set theory. Finally, a fusion image is created according to the classified results. Experiments indicate that the proposed method is superior to traditional algorithms with good anti-noise performance.