Perceptual image quality: A dissimilarity measure to quantify the degradation of image quality

摘要

This paper introduces a novel image quality metric that outperforms other existing and widely used metrics, namely the Root Mean Square Error (RMSE) and Structural Similarity Index (SSIM). Objective methods for assessing perceptual image quality are important for many image processing applications nowadays, however none of the existing metrics are completely satisfactory. Structural Similarity Index has been found promising in some cases, however for medical imaging (especially for CT and MRI images) it still remains immature. Perceptual Dissimilarity (PD) has been found reliable considering medical imaging demands; however for natural images, SSIM still outperforms PD. On that perspective, the proposed image quality metric named Perceptual Image Quality (PIQ) is more consistent with the human visual system's characteristics than SSIM and RMSE.