28 January 2008 Image quality evaluation using generalized natural image
Author Affiliations +
Abstract
Most image quality metrics are derived from feature values of specified test charts. However, such test charts can explain only a small portion of the comprehensive performances on image quality of imaging systems. Thus, designers of imaging systems need to check every possible type of natural image to verify the performance even if they check every image quality factor by test charts. But it is not clear how many and what types of images should be used to verify the performances. Meanwhile a number of studies have shown that the amplitude spectrum of natural images falls inversely with spatial frequency. This paper proposes a new image quality evaluation methodology using a quasi-random noise image that has 1/f spectrum property as a generalized natural image. After being processed by image processing operations, the power spectra of the image show reasonable responses to the operations and their parameters. In addition, a metric derived from this image can predict the subjective judgments on the spatial reproducibility of imaging systems with a high correlation coefficient. The results suggest that this image can be used for the purpose of evaluation of the comprehensive performance on image quality of imaging systems.
© (2008) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Kenji Kagitani "Image quality evaluation using generalized natural image", Proc. SPIE 6808, Image Quality and System Performance V, 68080S (28 January 2008); doi: 10.1117/12.765151; https://doi.org/10.1117/12.765151
PROCEEDINGS
8 PAGES


SHARE
Advertisement
Advertisement
Back to Top