Multi-focus image fusion based on non-negative matrix factorization and difference images

• An image fusion method via non-negative matrix factorization and difference images is proposed.
• Temporary fused image is constructed by using non-negative matrix factorization.
• Difference image׳s salient features are used to detect the source image׳s focused regions.
• The new fusion method outperforms the other existing image fusion approaches.

Multi-focus image fusion means to fuse multiple source images with different focus settings into one image, so that the resulting image appears sharper. One of the keys to image fusion is how to represent the source images effectively and completely. To address this problem, in this study a novel multi-focus image fusion scheme based on non-negative matrix factorization (NMF) and difference images is proposed. The temporary fused image is constructed by fusing the registered source images with NMF. The focused regions of the source images are detected by the salient features of the difference images between the temporary fused image and source images. The final fused image is produced by combining the focused regions. The experimental results demonstrate that the proposed method is capable of efficiently representing the source images and significantly improving the fusion quality compared to the other existing fusion methods, in terms of visual and quantitative evaluations.