Keywords

Natural scenes, optical environment, shadows, spatial cone-excitation ratios, color constancy

Abstract

In the natural world, the spectrum and geometry of the illumination from the sun and sky vary over the day. These geometric changes make it especially difficult to extract from the reflected light invariant signals for surface color perception. The aim of this study was to test the utility of certain combinations of retinal cone excitations, in particular, spatial cone-excitation ratios, known to be approximately invariant under non-geometric changes in illumination. Hyperspectral radiance images were acquired at roughly hourly intervals in four outdoor scenes in the Minho region of Portugal. Spatial resolution of the camera was 1344×1024 pixels. Peak-transmission wavelength increased in 10-nm steps from 400 to 720 nm. From these time-lapse hyperspectral sequences, spatial cone-excitation ratios were estimated at 10,000 randomly sampled pairs of points in each scene. Mean relative deviations in ratios were, as expected, generally large, but when sampling was limited to nearby points or moderate time intervals, they were much smaller, below threshold for detecting violations in ratio invariance. A similar result was obtained with spatial ratios of linear, opponent-color and cone-opponent combinations of cone excitations. A previously reported dependence between differences in logarithmic, opponent-color combinations of cone excitations was also found to be preserved over time intervals. In a changing natural world, stable surface color perception may be supported by several approximately invariant signals derived from cone excitations.

Start Date

15-5-2015 9:25 AM

End Date

15-5-2015 9:50 AM

Session Number

05

Session Title

Early Vision

Share

COinS
 
May 15th, 9:25 AM May 15th, 9:50 AM

Time-Lapse Statistics of Cone Signals from Natural Scenes

In the natural world, the spectrum and geometry of the illumination from the sun and sky vary over the day. These geometric changes make it especially difficult to extract from the reflected light invariant signals for surface color perception. The aim of this study was to test the utility of certain combinations of retinal cone excitations, in particular, spatial cone-excitation ratios, known to be approximately invariant under non-geometric changes in illumination. Hyperspectral radiance images were acquired at roughly hourly intervals in four outdoor scenes in the Minho region of Portugal. Spatial resolution of the camera was 1344×1024 pixels. Peak-transmission wavelength increased in 10-nm steps from 400 to 720 nm. From these time-lapse hyperspectral sequences, spatial cone-excitation ratios were estimated at 10,000 randomly sampled pairs of points in each scene. Mean relative deviations in ratios were, as expected, generally large, but when sampling was limited to nearby points or moderate time intervals, they were much smaller, below threshold for detecting violations in ratio invariance. A similar result was obtained with spatial ratios of linear, opponent-color and cone-opponent combinations of cone excitations. A previously reported dependence between differences in logarithmic, opponent-color combinations of cone excitations was also found to be preserved over time intervals. In a changing natural world, stable surface color perception may be supported by several approximately invariant signals derived from cone excitations.