A Multi-focus Image-fusion Scheme Based on non-Uniform Triangular Partition

This paper presents a spatial-domain multi-focus image fusion algorithm based on Non-uniform Triangular Partition (NTP). Generally, in the same region of two images with different focus, the number and size of segmented grids in the sub-image with clearer focus will be more and smaller than those in the non-focused fuzzy sub-image. Therefore, fusion pixels can be judged and marked in the Fusion Guiding Map (FGM) by the smaller partitioned triangle-size of the same region in two different-focus images. On the other hand, different control error leads to different segmentation effects in NTP, a set of Control Error is used to obtain different FGMs which are superimposed to form the final FGM that will be further optimized by morphological operation. Finally, each fused pixel is selected from the multi-focus images according to the optimized FGM to form the final fused image. The experimental results indicate that the subjective visual effect and objective index of the NTP-based fusion image are better than those from other frequency-domain methods.