Determining Visibility Thresholds for Spatial and Spatiotemporal Chromatic Noise
As imaging technology continues to advance toward higher resolutions, changes in sensor design are increasing the likelihood of noise manifesting within an image. Denoising algorithms are used when needed; however, they may be computationally intensive and time expensive. A previous work has shown that denoising algorithms for still images are not necessarily appropriate for temporal sequences. Research into the visibility of noise within images, either still or moving, is necessary to improve future image processing algorithms. Two types of noise can occur within an image as it relates to human vision: achromatic and chromatic noise. As an extension of a previous study on achromatic noise perception, we investigate the perception of chromatic noise in the perceptually uniform and human vision-based IPT color space. Noise was investigated in three color variations: a red-green variant, a blue-yellow variant, and a combination of the two in equal contributions. Each color variation was also examined in two imaging situations, i.e., one static pattern and one temporal sequence. Results show that all conditions exhibit low-pass characteristics, where the sensitivity to high-frequency noise patterns quickly declines in comparison to low and middle frequencies. Sensitivity is higher for low frequencies in a temporal condition versus a static condition, but this difference is small and decreases as we approach high frequencies. Finally, there does not appear to be a significant dependence on the color channel. Research into the visibility of noise within images, either still or moving, is necessary to improve future image processing algorithms.
- Print ISSN
- 1545-0279
- Electronic ISSN
- 2160-2492
- Published
- 2019-03
- Content type
- Original Research
- Keywords
- Chromatic noise, color vision, contrast sensitivity, perception, visibility threshold
- DOI
- 10.5594/JMI.2018.2887289