Home > Research > Publications & Outputs > NoRM
View graph of relations

NoRM: no-reference image quality metric for realistic image synthesis

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Published

Standard

NoRM: no-reference image quality metric for realistic image synthesis. / Herzog, Robert; Cadik, Martin; Aydin, Tunc Ozan et al.
In: Computer Graphics Forum, Vol. 31, No. 2 Pt 3, 05.2012, p. 545-554.

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Harvard

Herzog, R, Cadik, M, Aydin, TO, Kim, KI, Myszkowski, K & Seidel, H-P 2012, 'NoRM: no-reference image quality metric for realistic image synthesis', Computer Graphics Forum, vol. 31, no. 2 Pt 3, pp. 545-554. https://doi.org/10.1111/j.1467-8659.2012.03055.x

APA

Herzog, R., Cadik, M., Aydin, T. O., Kim, K. I., Myszkowski, K., & Seidel, H-P. (2012). NoRM: no-reference image quality metric for realistic image synthesis. Computer Graphics Forum, 31(2 Pt 3), 545-554. https://doi.org/10.1111/j.1467-8659.2012.03055.x

Vancouver

Herzog R, Cadik M, Aydin TO, Kim KI, Myszkowski K, Seidel H-P. NoRM: no-reference image quality metric for realistic image synthesis. Computer Graphics Forum. 2012 May;31(2 Pt 3):545-554. doi: 10.1111/j.1467-8659.2012.03055.x

Author

Herzog, Robert ; Cadik, Martin ; Aydin, Tunc Ozan et al. / NoRM : no-reference image quality metric for realistic image synthesis. In: Computer Graphics Forum. 2012 ; Vol. 31, No. 2 Pt 3. pp. 545-554.

Bibtex

@article{130956e0617d49198d11688c53b8f081,
title = "NoRM: no-reference image quality metric for realistic image synthesis",
abstract = "Synthetically generating images and video frames of complex 3D scenes using some photo-realistic rendering software is often prone to artifacts and requires expert knowledge to tune the parameters. The manual work required for detecting and preventing artifacts can be automated through objective quality evaluation of synthetic images. Most practical objective quality assessment methods of natural images rely on a ground-truth reference, which is often not available in rendering applications. While general purpose no-reference image quality assessment is a difficult problem, we show in a subjective study that the performance of a dedicated no-reference metric as presented in this paper can match the state-of-the-art metrics that do require a reference. This level of predictive power is achieved exploiting information about the underlying synthetic scene (e.g., 3D surfaces, textures) instead of merely considering color, and training our learning framework with typical rendering artifacts. We show that our method successfully detects various non-trivial types of artifacts such as noise and clamping bias due to insufficient virtual point light sources, and shadow map discretization artifacts. We also briefly discuss an inpainting method for automatic correction of detected artifacts.",
keywords = "I.3.3 [Computer Graphics], Picture/Image Generation—Image Quality Assessment",
author = "Robert Herzog and Martin Cadik and Aydin, {Tunc Ozan} and Kim, {Kwang In} and Karol Myszkowski and Hans-Peter Seidel",
year = "2012",
month = may,
doi = "10.1111/j.1467-8659.2012.03055.x",
language = "English",
volume = "31",
pages = "545--554",
journal = "Computer Graphics Forum",
issn = "1467-8659",
publisher = "Wiley-Blackwell",
number = "2 Pt 3",

}

RIS

TY - JOUR

T1 - NoRM

T2 - no-reference image quality metric for realistic image synthesis

AU - Herzog, Robert

AU - Cadik, Martin

AU - Aydin, Tunc Ozan

AU - Kim, Kwang In

AU - Myszkowski, Karol

AU - Seidel, Hans-Peter

PY - 2012/5

Y1 - 2012/5

N2 - Synthetically generating images and video frames of complex 3D scenes using some photo-realistic rendering software is often prone to artifacts and requires expert knowledge to tune the parameters. The manual work required for detecting and preventing artifacts can be automated through objective quality evaluation of synthetic images. Most practical objective quality assessment methods of natural images rely on a ground-truth reference, which is often not available in rendering applications. While general purpose no-reference image quality assessment is a difficult problem, we show in a subjective study that the performance of a dedicated no-reference metric as presented in this paper can match the state-of-the-art metrics that do require a reference. This level of predictive power is achieved exploiting information about the underlying synthetic scene (e.g., 3D surfaces, textures) instead of merely considering color, and training our learning framework with typical rendering artifacts. We show that our method successfully detects various non-trivial types of artifacts such as noise and clamping bias due to insufficient virtual point light sources, and shadow map discretization artifacts. We also briefly discuss an inpainting method for automatic correction of detected artifacts.

AB - Synthetically generating images and video frames of complex 3D scenes using some photo-realistic rendering software is often prone to artifacts and requires expert knowledge to tune the parameters. The manual work required for detecting and preventing artifacts can be automated through objective quality evaluation of synthetic images. Most practical objective quality assessment methods of natural images rely on a ground-truth reference, which is often not available in rendering applications. While general purpose no-reference image quality assessment is a difficult problem, we show in a subjective study that the performance of a dedicated no-reference metric as presented in this paper can match the state-of-the-art metrics that do require a reference. This level of predictive power is achieved exploiting information about the underlying synthetic scene (e.g., 3D surfaces, textures) instead of merely considering color, and training our learning framework with typical rendering artifacts. We show that our method successfully detects various non-trivial types of artifacts such as noise and clamping bias due to insufficient virtual point light sources, and shadow map discretization artifacts. We also briefly discuss an inpainting method for automatic correction of detected artifacts.

KW - I.3.3 [Computer Graphics]

KW - Picture/Image Generation—Image Quality Assessment

U2 - 10.1111/j.1467-8659.2012.03055.x

DO - 10.1111/j.1467-8659.2012.03055.x

M3 - Journal article

VL - 31

SP - 545

EP - 554

JO - Computer Graphics Forum

JF - Computer Graphics Forum

SN - 1467-8659

IS - 2 Pt 3

ER -