Multifocus image fusion using region segmentation and spatial frequency

摘要

Image fusion is a process of combining complementary information from multiple images of the same scene into an image, so that the resultant image contains a more accurate description of the scene than any of the individual source images. In this paper, a new region-based multifocus image fusion method is proposed. The motivation of our proposed method lies in the fact that region-based image fusion methods could be more meaningful than pixel-based fusion methods which just consider individual pixels or associated local neighborhoods of pixels in the fusion process. The fusion process contains the following steps: firstly, multifocus images are fused using the simple average method. Then the intermediate fused image is segmented using the normalized cut method. Then the two source images are segmented according to the segmenting result of the intermediate fused image. Finally, the corresponding segmented regions of the source images are fused according to their spatial frequencies. Experimental results on several pairs of multi-focus image show that the proposed method can give good results. The proposed method is more robust to misregistration or slight motion of the object than the pixel-based method.