Multi-Focus Image Fusion Based on Progressive Pixel Extraction and Morphology

This paper proposes a novel method that addresses the problem of artifacts, block effects and wrong extractions in local non-registered regions of the existing multi-focus image fusion methods.The proposed method involves taking the strong edges as the extraction basis, and adopting the sum of all high frequency coefficients' absolute values in a neighborhood of the extracted location as the extraction measure.This method employs a progressive strategy to extract sharp pixels in accordance with a series of continually increased neighborhoods.As a result, the contradiction between precisely extracting sharp pixels and extracting all pixels is resolved.Moreover, the proposed method introduces the '8-connected objects' morphology method to distinguish the extracted areas.Two groups of experimental results of fusing multi-focus images, with and without local non-registered regions, demonstrate that the proposed method solves the difficult problem of fusing the local non-registered regions, and is superior to the existing methods in terms of visual effects and objective evaluation on registered source images fusion.