Effect of scene complexity on colour constancy with real three-dimensional scenes and objects

The effect of scene complexity on colour constancy was tested with a novel technique in which a virtual image of a real 3-D test object was projected into a real 3-D scene. Observers made discriminations between illuminant and material changes in simple and complex scenes. The extent of colour constancy achieved varied little with either scene structure or test-object colour, suggesting a dominant role of local cues in determining surface-colour judgments.     

Surface-illuminant ambiguity and color constancy: effects of scene complexity and depth cues

Two experiments were conducted to study how scene complexity and cues to depth affect human color constancy. Specifically, two levels of scene complexity were compared. The low-complexity scene contained two walls with the same surface reflectance and a test patch which provided no information about the illuminant. In addition to the surfaces visible in the low-complexity scene, the high-complexity scene contained two rectangular solid objects and 24 paper samples with diverse surface reflectances. Observers viewed illuminated objects in an experimental chamber and adjusted the test patch until it appeared achromatic. Achromatic settings made tinder two different illuminants were used to compute an index that quantified the degree of constancy. Two experiments were conducted: one in which observers viewed the stimuli directly, and one in which they viewed the scenes through an optical system that reduced cues to depth. In each experiment, constancy was assessed for two conditions. In the valid-cue condition, many cues provided valid information about the illuminant change. In the invalid-cue condition, some image cues provided invalid information. Four broad conclusions are drawn from the data: (a) constancy is generally better in the valid-cue condition than in the invalid-cue condition: (b) for the stimulus configuration used, increasing image complexity has little effect in the valid-cue condition but leads to increased constancy in the invalid-cue condition; (c) for the stimulus configuration used, reducing cues to depth has little effect for either constancy condition: and (d) there is moderate individual variation in the degree of constancy exhibited, particularly in the degree to which the complexity manipulation affects performance.     

Colour constancy in goldfish and man: influence of surround size and lightness

Colour constancy was investigated by using a series of 10 simultaneously presented surface colours ranging in small steps from green through gray to red-purple. Goldfish were trained to select one medium test field when the entire setup was illuminated with white light. In the tests, either red or green illumination was used. Colour constancy, as inferred from the choice behaviour, was perfect under green illumination when the test fields were presented on a gray or a white background, but imperfect on a black background. Under red illumination and a white background, however, colour constancy was overcompensated. Here, a colour contrast effect was observed. The influence of background lightness was also found when the surround was restricted to a narrow annulus of 4-11 mm width (test field diameter: 14 mm). By applying colour metrics it could be shown that the von Kries coefficient law can describe the overall effect of colour constancy. For an explanation of the effect of surround size and lightness, lateral inhibitory interactions have to be assumed in addition, which are also responsible for simultaneous colour contrast. Very similar results were obtained in experiments with the same colours in human subjects. They had to name the test field appearing 'neutral' under the different illumination and surround conditions, as tested in the goldfish experiment.     

Perceived depth of 3-D objects in 3-D scenes

Effects of information specifying the position of an object in a 3-D scene were investigated in two experiments with twelve observers. To separate the effects of the change in scene position from the changes in the projection that occur with increased distance from the observer, the same projections were produced by simulating (a) a constant object at different scene positions and (b) different objects at the same scene position. The simulated scene consisted of a ground plane, a ceiling plane, and a cylinder on a pole attached to both planes. Motion-parallax scenes were studied in one experiment; texture-gradient scenes were studied in the other. Observers adjusted a line to match the perceived internal depth of the cylinder. Judged depth for objects matched in simulated size decreased as simulated distance from the observer increased. Judged depth decreased at a faster rate for the same projections shown at a constant scene position. Adding object-centered depth information (object rotation) increased judged depth for the motion-parallax displays. These results demonstrate that the judged internal depth of an object is reduced by the change in projection that occurs with increased distance, but this effect is diminished if information for change in scene position accompanies the change in projection.     

Illumination change at a depth edge can reduce lightness constancy

Lightness constancy requires that a surface retain its lightness not only when the illumination is changed but also when the surface is moved from one background to another. Occlusion of one surface by another frequently results in a retinal juxtaposition of patches under different illuminations. At such edges, retinal luminance ratios can be much higher than in scenes with a single illumination. We demonstrate that such retinal adjacencies can produce failures of lightness constancy. We argue that they are responsible for departures from perfect lightness constancy in two prior experiments that examined the effects of depth relations on lightness constancy.     

The perception of lightness in 3-D curved objects

Lightness constancy in complex scenes requires that the visual system take account of information concerning variations of illumination falling on visible surfaces. Three experiments on the perception of lightness for three-dimensional (3-D) curved objects show that human observers are better able to perform this accounting for certain scenes than for others. The experiments investigate the effect of object curvature, illumination direction, and object shape on lightness perception. Lightness constancy was quite good when a rich local gray-level context was provided. Deviations occurred when both illumination and reflectance changed along the surface of the objects. Does the perception of a 3-D surface and illuminant layout help calibrate lightness judgments? Our results showed a small but consistent improvement between lightness matches on ellipsoid shapes, relative to flat rectangle shapes, under illumination conditions that produce similar image gradients. Illumination change over 3-D forms is therefore taken into account in lightness perception.     

Cone contributions to colour constancy

Colour constancy refers to the stable perception of object colour under changing illumination conditions. This problem has been reformulated as relational colour constancy, or the ability of the observer to discriminate between material changes and changes in illumination. It has been suggested that local cone excitation ratios play a prominent role in achieving such constancy. Here we show that perceptual colour constancy measured by achromatic adjustments is to a large part complete after 25 ms. This speaks against a prominent role for receptor adaptation, which takes significantly longer. We also found no difference in colour constancy between colour changes that were compatible with a change of illuminant, and between colour changes where local cone ratios were uncorrelated between the two illuminants. Our results show that constant cone ratios are not necessary for colour constancy.     

Developmental changes in colour constancy in a naturalistic object selection task

When the illumination falling on a surface change, so does the reflected light. Despite this, adult observers are good at perceiving surfaces as relatively unchanging—an ability termed colour constancy. Very few studies have investigated colour constancy in infants, and even fewer in children. Here we asked whether there is a difference in colour constancy between children and adults; what the developmental trajectory is between six and 11 years; and whether the pattern of constancy across illuminations and reflectances differs between adults and children. To this end, we developed a novel, child-friendly computer-based object selection task. In this, observers saw a dragon's favourite sweet under a neutral illumination and picked the matching sweet from an array of eight seen under a different illumination (blue, yellow, red, or green). This set contained a reflectance match (colour constant; perfect performance) and a tristimulus match (colour inconstant). We ran two experiments, with two-dimensional scenes in one and three-dimensional renderings in the other. Twenty-six adults and 33 children took part in the first experiment; 26 adults and 40 children took part in the second. Children performed better than adults on this task, and their performance decreased with age in both experiments. We found differences across illuminations and sweets, but a similar pattern across both age groups. This unexpected finding might reflect a real decrease in colour constancy from childhood to adulthood, explained by developmental changes in the perceptual and cognitive mechanisms underpinning colour constancy, or differences in task strategies between children and adults.

Highlights

• Six- to 11-year-old children demonstrated better performance than adults on a colour constancy object selection task.
• Performance decreased with age over childhood.
• These findings may indicate development of cognitive strategies used to overcome automatic colour constancy mechanisms.

     

Lightness constancy: a direct test of the illumination-estimation hypothesis

Many models of color constancy assume that the visual system estimates the scene illuminant and uses this estimate to determine an object's color appearance. A version of this illumination-estimation hypothesis, in which the illuminant estimate is associated with the explicitly perceived illuminant, was tested. Observers made appearance matches between two experimental chambers. Observers adjusted the illumination in one chamber to match that in the other and then adjusted a test patch in one chamber to match the surface lightness of a patch in the other. The illumination-estimation hypothesis, as formulated here, predicted that after both matches the luminances of the light reflected from the test patches would be identical. The data contradict this prediction. A second experiment showed that manipulating the immediate surround of a test patch can affect perceived lightness without affecting perceived illumination. This finding also falsifies the illumination-estimation hypothesis.     

The influence of chromatic and achromatic variability on chromatic induction and perceived colour

Judgments of the colour of a surface are influenced by the colour of the surrounding. To determine whether only the average colour of the surrounding matters, or also the chromatic variability, judgments in colourful scenes are often compared with ones in which a target is surrounded by a plain background that provides the same average physical illumination of the retina as the colourful scene. The variability sometimes makes a difference (eg Shevell and Wei, 1998 Vision Research 38 1561-1566), and sometimes it does not (eg Brenner and Cornelissen, 1998 Vision Research 38 1789-1793). Is this because of the nonlinearity in cone responses? We designed scenes that stimulated the cones in an equivalent manner, both on average and in terms of variability, and yet differed markedly in chromatic variability. The more colourful surroundings had considerably less influence on subjects' colour judgments. We conclude that early cone-specific regulation of sensitivity cannot be responsible for the change in perceived colour, and deduce that chromatic induction takes place after contrast gain control.     

The AMBEGUJAS phenomenon and colour constancy

The AMBEGUJAS phenomenon is a reversible flat figure that is spontaneously shifting between two apparent 3-D shapes-'tile' and 'roof'. 2-D perceptions have very rarely been reported. Tied to the shifts between the tile and roof shapes are remarkable changes of perceived colour. In our example, the tile appears to have orange (top half) and blue-green (bottom half) surface colours in white light. The roof appears grey but in an orange illumination and with a blue-green shadow. This phenomenon appears whether a grey display is presented in two coloured illuminations, or a chromatic display with two surface colours (orange and blue-green) is presented in white light. In the coloured illuminations the tile is an example of non-constancy, since its colours are non-veridical colour perceptions. The centre stripe of the display appears to have the same orange and blue-green colours as the lateral stripes but in a shadow. This seems like a colour constancy in a non-constancy situation. An alternative to the classical definition of colour constancy is discussed.     

Information limits on identification of natural surfaces by apparent colour

By adaptational and other mechanisms, the visual system can compensate for moderate changes in the colour of the illumination on a scene. Although the colours of most surfaces are perceived to be constant ('colour constancy'), some are not. The effect of these residual colour changes on the ability of observers to identify surfaces by their apparent colour was determined theoretically from high-resolution hyperspectral images of natural scenes under different daylights with correlated colour temperatures 4,300 K, 6,500 K, and 25,000 K. Perceived differences between colours were estimated with an approximately uniform colour-distance measure. The information preserved under illuminant changes increased with the number of surfaces in the sample, but was limited to a relatively low asymptotic value, indicating the importance of physical factors in constraining identification by apparent colour.     

Scene articulation: dependence of illuminant estimates on number of surfaces

The ability of observers to detect changes in illuminant over two scenes containing different random samples of reflecting surfaces was determined in an experiment with Mondrian-like patterns containing different numbers of coloured patches. Performance was found to improve as the number of patches increased from 9 to 49. In principle, observers could have used space-average scene colour as the cue ('grey-world' hypothesis) or the colour of the brightest surface in the scene ('bright-is-white' hypothesis), as the two cues generally covary. In a second experiment, observers matched illuminants across different patterns in which the space-average cue and the brightest-patch cue were independently manipulated. The articulation of the patterns was varied: the number of patches increased from 49 (patch width 1 deg visual angle) to over 50000 (patch width 0.03 deg), while the gamut of colours was held constant. Space-average colour was found to be the dominant cue with all patterns except for those with the largest patches.     

A test of size-scaling and relative-size hypotheses for the moon illusion

In two experiments participants reproduced the size of the moon in pictorial scenes under two conditions: when the scene element was normally oriented, producing a depth gradient like a floor, or when the scene element was inverted, producing a depth gradient like a ceiling. Target moons were located near to or far from the scene element. Consistent with size constancy scaling, the illusion reversed when the "floor" of a pictorial scene was inverted to represent a "ceiling." Relative size contrast predicted a reduction or increase in the illusion with no change in direction. The relation between pictorial and natural moon illusions is discussed.     

PSYCHOPHYSICAL INVARIANTS OF ACHROMATIC COLOUR VISION: III. Colour constancy and its relation to identification of illumination

L ie , I. Psychophysical invariants of achromatic colour vision. III. Colour constancy and its relation to identification of illumination. Scand. J. Psychol ., 1969, 10 , 7.69–281.—Identification of illumination as a separate determinant of the dimensions of achromatic colour was investigated with classical methods for manipulating shifts in perceived level of illumination, while controlling the changes in colour contrast brought about by these methods. With the spot-shadow method, constancy effects may be explained by laws of colour contrast, but not by perceived shifts in illumination, whereas neither explanation holds with the cast-shadow method. Results with a method of conceiled cast-shadow may be consistent with both interpretations.     

PSYCHOPHYSICAL INVARIANTS OF ACHROMATIC COLOUR VISION: II. Albedo/illumination substitution

Lie , I. Psychophysical invariants of achromatic colour vision. II. Albedo/ illumination substitution. Scand. J. Psychol., 1969,10, 176–184.—Can ‘identification of illumination’ be identified as one of the determinants of achromatic vision? Colour constancy is redefined in terms of deviation from albedo/ illumination substitution (A/I-substitution). A series of A/I-substitution experiments, performed under conditions assumed to provide minimal possibilities for identification of illumination, indicate that complete A/I-substitution (zero constancy) is obtained for both brightness and whiteness, provided the test fields are equated with respect to colour temperature and perceived texture.     

Lightness of an object under two illumination levels

Anchoring theory (Gilchrist et al, 1999 Psychological Review 106 795-834) predicts a wide range of lightness errors, including failures of constancy in multi-illumination scenes and a long list of well-known lightness illusions seen under homogeneous illumination. Lightness values are computed both locally and globally and then averaged together. Local values are computed within a given region of homogeneous illumination. Thus, for an object that extends through two different illumination levels, anchoring theory produces two values, one for the patch in brighter illumination and one for the patch in dimmer illumination. Observers can give matches for these patches separately, but they can also give a single match for the whole object. Anchoring theory in its current form is unable to predict these object matches. We report eight experiments in which we studied the relationship between patch matches and object matches. The results show that the object match represents a compromise between the match for the patch in the field of highest illumination and the patch in the largest field of illumination. These two principles are parallel to the rules found for anchoring lightness: highest luminance rule and area rule.     

Surface formation and depth in monocular scene perception

The visual perception of monocular stimuli perceived as 3-D objects has received considerable attention from researchers in human and machine vision. However, most previous research has focused on how individual 3-D objects are perceived. Here this is extended to a study of how the structure of 3-D scenes containing multiple, possibly disconnected objects and features is perceived. Da Vinci stereopsis, stereo capture, and other surface formation and interpolation phenomena in stereopsis and structure-from-motion suggest that small features having ambiguous depth may be assigned depth by interpolation with features having unambiguous depth. I investigated whether vision may use similar mechanisms to assign relative depth to multiple objects and features in sparse monocular images, such as line drawings, especially when other depth cues are absent. I propose that vision tends to organize disconnected objects and features into common surfaces to construct 3-D-scene interpretations. Interpolations that are too weak to generate a visible surface percept may still be strong enough to assign relative depth to objects within a scene. When there exists more than one possible surface interpolation in a scene, the visual system's preference for one interpolation over another seems to be influenced by a number of factors, including: (i) proximity, (ii) smoothness, (iii) a preference for roughly frontoparallel surfaces and 'ground' surfaces, (iv) attention and fixation, and (v) higher-level factors. I present a variety of demonstrations and an experiment to support this surface-formation hypothesis.     

Information about three-dimensional shape and direction of illumination in a square-wave grating

A homogeneous grey picture and a 'Mondrian' type of picture were illuminated by a projector with square-wave gratings of thirty different contrast values used as slides. Ten observers reported whether the picture appeared three-dimensional (3-D) (pleated) or flat. 3-D responses in this situation indicate colour constancy 'at the cost of' nonveridical depth perception. The frequencies of 3-D responses were significantly higher for the structured picture than for the homogeneous grey one. In reports of the direction from which the apparent 3-D object appeared to be illuminated there was a significant preference for responses "from above" when the grating was horizontally oriented. With vertical orientation there was no preference for "from the left" or "from the right". The results from the first experiment contradict traditional cue theories of depth perception since the projection of the borders between the fields of the structured picture was invariant and expected to inform about the flatness of the picture. They are, however, in line with a model for perceptual analysis of reflected light into common and relative components proposed earlier by Bergstr?m. The difference in perceived direction of illumination between horizontally and vertically orientated gratings is discussed in connection with human ecology.     

Effects of collimation on perceived layout in 3-D scenes.

The effect of varying information for overall depth in a simulated 3-D scene on the perceived layout of objects in the scene was investigated in two experiments. Subjects were presented with displays simulating textured surfaces receded in depth. Pairs of markers were positioned at equal intervals within the scenes. The subject's task was to judge the depth between the intervals. Overall scene depth was varied by viewing through either a collimating lens or a glass disk. Judged depth for equal depth intervals decreased with increasing distance of the interval from the front of the scene. Judged depth was greater for collimated than for non-collimated viewing. Interestingly, collimated viewing resulted in a uniform rescaling of the perceived depth intervals.