AI Summary & Key Takeaways
AI InsightThis conclusion attests that using the CQM together with the traditional PSNR approach provides distinguished results.
📌 3 Key Takeaways
Abstract
Various methods for measuring perceptual image quality attempt to quantify the visibility of differences between an original digital image and its distorted version using a variety of known properties of the human vision system (HVS). In this paper, we propose a simple and effective full-reference color image quality measure (CQM) based on reversible luminance and chrominance (YUV) color transformation and peak signal-to-noise ratio (PSNR) measure. The main motivation of this new measure relies on a unique feature of the human eye response to the luminance and color. Experimental studies about the applicability of the CQM on a well-known test image under 6 different distortions, both perceivable by the human vision system and with the same PSNR value (i.e. 27.67), are presented. The CQM results are obtained as 39.56, 38.93, 38.08, 37.43, 37.10, and 36.79 dB for each distorted image, showing that image quality of the first image is noticeably higher than the others with respect to the same PSNR value. This conclusion attests that using the CQM together with the traditional PSNR approach provides distinguished results.
Özet
Various methods for measuring perceptual image quality attempt to quantify the visibility of differences between an original digital image and its distorted version using a variety of known properties of the human vision system (HVS). In this paper, we propose a simple and effective full-reference color image quality measure (CQM) based on reversible luminance and chrominance (YUV) color transformation and peak signal-to-noise ratio (PSNR) measure. The main motivation of this new measure relies on a unique feature of the human eye response to the luminance and color. Experimental studies about the applicability of the CQM on a well-known test image under 6 different distortions, both perceivable by the human vision system and with the same PSNR value (i.e. 27.67), are presented. The CQM results are obtained as 39.56, 38.93, 38.08, 37.43, 37.10, and 36.79 dB for each distorted image, showing that image quality of the first image is noticeably higher than the others with respect to the same PSNR value. This conclusion attests that using the CQM together with the traditional PSNR approach provides distinguished results.