SPIE Membership Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:


Print PageEmail PageView PDF

Illumination & Displays

Changes in image size affect perceived contrast

Results from contrast matching of noise pattern images have implications for the restoration and digital reproduction of cultural artifacts.
22 May 2007, SPIE Newsroom. DOI: 10.1117/2.1200704.0724

Original works of art and their reproductions are usually viewed under quite different conditions. Visual angle and physical size are two of several factors affecting color perception. Variations in size lead to changes in perceived contrast. In this research, a projector (the larger ’original’) and an LCD monitor (the smaller ’reproduction’) were used to reproduce achromatic noise patterns with very different physical sizes.1,2 These patterns consisted of random dots that were filtered to contain only a narrow band of spatial frequencies. The two devices were characterized so that the colors of the two displays matched. A matching technique was then used to collect data to develop a multiscale model for rendering images with very different sizes so that the apparent contrast remains constant.

The sensitivity of the human visual system to differences in light and dark is described by a function that relates the reciprocal of the contrast threshold (the minimum at which a sinusoidal pattern is just visible) to spatial frequency.3 This function is characterized as band-pass—i.e., tuned to a range of spatial frequencies—and shows that sensitivity is highest for patterns with a frequency of about three to four cycles per degree. These mechanisms change with the level of illumination, which is reflected in alteration of sensitivity at different light levels.4 In the approach of Maffei and Fiorentini,5 the visual cortex is considered as a spatial frequency analyzer. Using band-limited noise patterns of different sizes, we can study the effect of the physical size of an image on the perceived contrast in the corresponding frequency band.

Three one-octave cosine filters were used to filter a uniform ‘white’ noise pattern. The frequency responses of the filters are shown in Figure 1. Cosine log filters are symmetric on a log frequency axis and are a good approximation for Gaussian filters. Furthermore, an image filtered by a bank of cosine log filters can be reconstructed from its segments by simple addition.6 For three different visual angles and three levels of contrast and luminance, a total of 63 noise pattern images were rendered for both displays using the cosine log filters. In a darkened room, 14 observers adjusted the mean luminance level and contrast of the images on a projector screen to match the images displayed on an LCD monitor. Figures 2 and 3 present the means and corresponding 95% confidence intervals for the observers' matches for the adjusted contrast and mean luminance, respectively. The abscissa represents the relative magnifications of the monitor image compared with the image on the screen.

Figure 1. Frequency responses of octave filters centered at 0.5, 2, and 8 cycles per degree (cpd) of visual angle. For frequencies lower than 0.5cpd and higher than 8cpd, filter responses were set to unity for filters centered at 0.5 and 8cpd, respectively.

Figure 2. Contrast adjustment of digital light processing (DLP) images versus LCD images for 14 observers. Panels A, D, and G correspond to low-frequency noise patterns, centered at 0.5cpd, and CIE (International Commission on Illumination) L* values of 75, 50, and 25, respectively. Panels B, E, and H correspond to medium-frequency noise patterns. Panels C, F, and I correspond to high-frequency noise patterns, centered at 8cpd. Dashed lines in each panel show contrast of LCD images.

Figure 2 shows that for low spatial frequency patterns, observers had to increase the larger projector's image contrast to match that of the smaller image on the monitor. Conversely, high-spatial-frequency, small-sized images on the LCD were matched with images of lower contrast on the projector screen. This decrease in contrast was more pronounced for images with high mean luminance levels.

Compared with the mean luminance level of the LCD images, a decrease of the mean luminance level of the adjusted images was observed for all spatial frequency noise patterns, as shown in Figure 3. This decrease was more pronounced for smaller images at low and medium contrast. Future research will explore development of a multiscale model for both achromatic and chromatic images. This research will be applied to the reproduction of cultural heritage to improve the quality of appearance.

Figure 3. Luminance adjustment results for DLP images versus LCD images for 14 observers. Panels A, D, and G correspond to low-frequency noise patterns, centered at 0.5cpd, and three levels of contrast. Panels B, E, and H correspond to medium-frequency noise patterns, centered at 2cpd. Panels C, F, and I correspond to high-frequency noise patterns, centered at 8cpd. Dashed lines in each panel show mean luminance of LCD images.

Mahdi Nezamabadi, Ethan D. Montag, Roy S. Berns  
Munsell Color Science Laboratory
Rochester Institute of Technology (RIT)
Rochester, NY

Mahdi Nezamabadi received his BS and MS degrees in textile engineering from Amirkabir University of Technology, Iran, in 1992 and 2001, respectively. From 1997 to 2002 he worked as a lab instructor and researcher in the Color Science Department at the same university. He is currently a PhD candidate in imaging science at the Munsell Color Science Laboratory of RIT.

Ethan D. Montag received his PhD in experimental psychology in 1991 from the University of California, San Diego, working in color vision. In 2000, he was appointed assistant professor at the Center for Imaging Science at RIT, where he pursues work in color science in the Munsell Color Science Laboratory. His current interests include image quality, color gamut mapping, color vision, color tolerance measurement and visualization. He is a member of the Association for Research in Vision and Ophthalmology, the Optical Society of America, the International Colour Vision Society, the Inter-Society Color Council, and the Society for Imaging Science and Technology.

Roy S. Berns is the Richard S. Hunter Professor in Color Science, Appearance, and Technology at the Munsell Color Science Laboratory and graduate coordinator of the color science degree programs within the Center for Imaging Science at Rochester Institute of Technology. During 2005, he joined the executive committee of the International Association of Colour.