Image Quality Assessment Using Variational Mode Weighted Index

Abstract

The prime objective of an image quality assessment model is to assess the image quality consistently based on the whole image information. However, human perception does not match with the same philosophy. Based on this hypothesis, subjective and objective measures are commonly used for measuring image quality. Moreover, subjective evaluation is a timeconsuming process so objective quality measures are used in the majority of image processing applications. In this paper, a new objective measure (variational mode quality index (VMQI)) is proposed to determine the quality of the image. Initially the reference image and the processed image are decomposed separately to four-level 2-D decomposition using the Variational Mode Decomposition (VMD). Then, the structural similarities between the four decomposed modes are evaluated using SSIM. Separate weighting values are computed for each SSIM based on either energy or entropy of each mode of the original image. Finally, the VMQI is evaluated using the weighting and the SSIM values. The performance of the proposed quality measure, VMQI, is evaluated using the publicly available image database. The experimental result reveals that the VMQI has a higher correlation coefficient (CC) 0.996 as compared to other existing measures such as SSIM and MSE which have correlation values of 0.9566 and 0.9715.