A Novel Region Segmentation-Based Multi-focus Image Fusion Model

Abstract

Multi-focus image fusion scheme integrates multiple input images to obtain a composite fused image. Many research works have been carried out since years and various image fusion approaches were developed. The main idea behind the image fusion is to generate a fused image with enhanced quality and containing more information than that of individual source images. Nowadays, these image fusion techniques are implemented in many applications to combine multi-focus image data into a single composite image. Image fusion models can be categorized into two ways, spatial based fusion and transform based fusion. Transform based fusion is performed in three steps, (1) In the first step, transform coefficients from the input images are turned into transform domain frequencies. (2) In the second step, by applying the fusion rule, these transform coefficients are combined. (3) Through the process of inverse transform on the combined correlated images, fused composite image is generated. In this paper, we have introduced a novel region segmentation based multi-focus image fusion model and implemented it. Proposed model was thoroughly studied, analyzed, and compared with different multi-focus fusion models. Experimental results prove that the proposed model has high computational accuracy in terms of image quality and less error rate compared to traditional models.