A. Radonji's scientific contributions

A central goal in the study of color appearance is to develop and validate models that predict object color appearance from a physical scene description. Ultimately, we seek models that apply for any stimulus, and particularly for stimuli typical of natural viewing. One approach is to study color appearance using real illuminated objects in quasi-natural arrangements. This approach has the advantage that the measurements are likely to capture what happens for natural viewing. It has the disadvantage that it is challenging to manipulate the stimuli parametrically in theoretically interesting ways. At the other extreme, one can choose simplified stimulus sets (e.g., spots of light on uniform backgrounds, or Mondrian configurations). This approach has the advantage that complete characterization of performance within the set may be possible, and one can hope that any principles developed will have general applicability. On the other hand, there is no a priori guarantee that what is learned will indeed be helpful for predicting what happens for real illuminated objects. Here we consider an intermediate choice, the use of physically-accurate graphics simulations. These offer the opportunity for precise stimulus specification and control; particularly interesting is the ability to manipulate explicitly distal (object and illuminant) rather than proximal (image) stimulus properties. They also allow for systematic introduction of complexities typical of natural stimuli, thus making it possible to ask what features of natural viewing affect performance and providing the potential to bridge between the study of simplified stimuli and the study of real illuminated objects. Meeting abstract presented at VSS 2016

We use color to select objects and decide how to interact with them. How color supports cross-illumination object selection, however, is not well understood. This is particularly true for naturalistic tasks. In our experiments, subjects performed a naturalistic task that relied on color. On each trial, the subjects saw three simulated scenes: the model, the source, and the workspace. The model contained four target blocks that differed in reflectance; their arrangement varied across trials. The source contained eight blocks — one pair of competitors for each target block, whose degree of color similarity relative to the target varied across trials. The subjects' task was to recreate the model arrangement of blocks in the workspace by selecting the blocks from the source. The model scene was rendered under a standard illuminant, while the source and workspace were rendered under a test illuminant. As the subjects copied the model, they selected blocks across an illuminant change. By analyzing subjects' choices across a series of pairwise combinations of competitors, we inferred a selection-based match for each target and illuminant change using a variant of maximum likelihood difference scaling. We then used these selection-based matches to compute selection-based color constancy indices (CCI), which could range from 0 (no constancy) to 1 (perfect constancy). Overall, constancy was fairly good in our naturalistic task (mean CCI = 0.46), but varied considerably across subjects (0.22–0.86). Our results show that the visual system's mechanisms of color constancy serve to support veridical object selection across changes in illumination.

The phenomenon of colour constancy in human visual perception keeps surface colours constant, despite changes in their reflected light due to changing illumination. Although colour constancy has evolved under a constrained subset of illuminations, it is unknown whether its underlying mechanisms, thought to involve multiple components from retina to cortex, are optimised for particular environmental variations. Here we demonstrate a new method for investigating colour constancy using illumination matching in real scenes which, unlike previous methods using surface matching and simulated scenes, allows testing of multiple, real illuminations. We use real scenes consisting of solid familiar or unfamiliar objects against uniform or variegated backgrounds and compare discrimination performance for typical illuminations from the daylight chromaticity locus (approximately blue-yellow) and atypical spectra from an orthogonal locus (approximately red-green, at correlated colour temperature 6700 K), all produced in real time by a 10-channel LED illuminator. We find that discrimination of illumination changes is poorer along the daylight locus than the atypical locus, and is poorest particularly for bluer illumination changes, demonstrating conversely that surface colour constancy is best for blue daylight illuminations. Illumination discrimination is also enhanced, and therefore colour constancy diminished, for uniform backgrounds, irrespective of the object type. These results are not explained by statistical properties of the scene signal changes at the retinal level. We conclude that high-level mechanisms of colour constancy are biased for the blue daylight illuminations and variegated backgrounds to which the human visual system has typically been exposed.

Color constancy is often studied using adjustment procedures. In real life, however, we rarely adjust object colors. Rather, we use color to identify and choose objects. We studied constancy using a novel paradigm, adapted from the blocks-copying task of Ballard et al (2005). In our experiment, subjects were asked to complete a fairly natural task that required them to judge object color across an illumination change. At the beginning of each trial the subject saw three rendered scenes —the target, the source and the test— presented on a computer display. The target scene contained four colored blocks of different simulated reflectance. Their arrangement varied randomly on each trial. The source scene contained eight blocks: one pair of potential matches for each target block. The degree of similarity of each potential match to the target varied across trials. The test scene contained four identical dark gray blocks. The subjects' task was to replace the gray blocks with blocks chosen from the source, so as to recreate the arrangement in the target scene as closely as possible. In the illuminant-constant condition, all three scenes were rendered under the same illumination (D65). In the illuminant-changed condition, the simulated illuminations of the source and test were changed to 12000°K. Based on the subjects' choices, we inferred their perceptual matches for each target block in each condition via a variant of the maximum likelihood difference scaling method. Two main findings were consistent across our four subjects: (1) When the illumination was constant the distance between the target block and its choice-based match was small (2.2 - 3.9 ∆E), supporting the validity of our method. (2) When the illumination changed, the choice-based matches indicated good constancy (constancy indices 0.7 - 0.8). Our results show that color constancy is high when probed using a natural task. Meeting abstract presented at VSS 2014