Image Quality Assessment Using the SSIM and the Just Noticeable Difference Paradigm

By Jeremy R. Flynn, Steve Ward, IV JulianAbich, David Poole. 2013

47 citations recorded by the source corpus.

The structural similarity index (SSIM) has been shown to be a superior objective image quality metric. A web-based pilot experiment was conducted with the goal of quantifying, through the use of a sample of human participants, a trend in SSIM values showing when the human visual system can begin to perceive distortions applied to reference images. The just noticeable difference paradigm was used to determine the point at which at least 50% of participants were unable to discern between compressed and uncompressed grayscale images. For four images, this point was at an SSIM value of 96, while for two images it was at 92, for an average of 95. These results suggest that, despite the wide differences in the type of image used, the point at which a human observer cannot determine that compression has been used hovers around an SSIM value of 95.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Andras Rozsa
2. Terrance E. Boult
3. Ethan M. Rudd
4. Pan He
5. Xiaolin Li
6. Qile Zhu
7. Xiaoyong Yuan
8. Manuel Günther

In Media Technology

1. Ethan M. Rudd
2. Xiaolin Li
3. Xiaoyong Yuan
4. Qile Zhu
5. Andras Rozsa
6. Terrance E. Boult
7. Pan He

Citations in the corpus

• Adversarial Examples: Attacks and Defenses for Deep Learning (2017, methodology): “Structural SIMilarity (SSIM) index [82] was adopted to measure the just noticeable difference of images.”
• Adversarial Diversity and Hard Positive Generation (2016, background): “Consistent with [7], this suggests that adversarial images should be quantified in terms of just noticeable difference.”