ENHANCED SATELLITE IMAGES FUSION ALGORITHM USING COMBINED BI-DIMENSIONAL EMPIRICAL MODE DECOMPOSITION AND SHEARLET TRANSFORM BASED ON QGA OPTIMIZATION

Abstract

The satellite images fusion is a process of merging data of the same scene from the panchromatic (PAN) and multispectral (MS) images captured by different instruments. This paper introduces a new satellite images fusion method using Bidimensional Empirical Mode Decomposition (BEMD) and shearlet Transform (ST). BEMD is an efficient method for satellite images processing due to its advantage of spectral data maintaining. Basically, it is used to transform the MS image into subsets named Intrinsic Mode Functions (IMFs). Hence these IMFs and the PAN image are divided by shearlet into high and low frequencies subsets, then the high frequency ones of IMFs are replaced with their corresponding subsets of the PAN image. At last, inverse shearlet and inverse BEMD are implemented to get the fused MS image. Shearlet is preferred due to its optimal representation of the anisotropic elements in the image and due to its capability in processing the continuity and digital data unlike curvelet and contourlet. The experiment results illustrated that the proposed method extracts more spatial details from PAN images with fewer losses in spectral quality of MS images compared to other classic fusion methods. For more enhancements, the fusion weights are estimated efficiently by a quantum genetic algorithm (QGA)-based approach. Keywords: Image, satellite, fusion, BEMD, Shearlet, QGA.