In order to efficiently represent source images and inhibit the blocking artifacts, a novel multi-focus image fusion scheme based on image decomposition and quad tree (QT) decomposition is proposed. The registered source images are first decomposed into cartoon and texture components by using image decomposition, and then, QT decomposition is performed on the cartoon and texture components. The salient features of the cartoon and texture components construct the feature space. The focused regions of the source images are detected by the salient features of the cartoon and texture components. The final fused image can be produced by combining the image regions that corresponding to the focused component regions. Experimental results have demonstrated that the proposed method can efficiently inhibit the blocking artifacts and significantly improve the fusion quality compared to the other existing fusion methods in both spatial and transform domain.