Morphology and Active Contour Model for Multi-Focus Image Fusion

Abstract

Multi-focus image fusion is an important technique that extracts sharpness regions from multiple images and composites them into a fully focused image. In this paper, a novel spatial domain based fusion algorithm for multi-focus image through gradient based decision map construction using morphology and active contour model is proposed. Firstly, the original focus maps are constructed based on the gradient of the images. Secondly, the coarse focus maps are produced by employing convolution on the original focus maps with a weighted kernel and then the final focus map is decided by the coarse focus maps. Thirdly, the original fusion decision map is obtained by processing the final focus map with morphological opening-and-closing operations and small object removing operation. Fourthly, the extracted boundaries are used as the initial value of the free boundary conditions active contour model and the final decision map is acquired according the original decision map and the boundary map. Finally, the fused image is produced based on the final decision map and the fusion rule. Experimental results show that the proposed algorithm performs better than the other five representative fusion algorithms in both the qualitative and quantitative evaluations.