Colour perception
What is colour? • Visible light forms a narrow band of frequencies in the electromagnetic spectrum. • Within this band, different frequencies (or wavelengths) have different hues, ranging from red (for long wavelength light) to violet (for short wave length light). • Paradoxically, it is this finding which severely hampered our understanding of colour vision until very recently.
Electromagnetic Spectrum
Colours of sunlight • Why is the sky blue, the sun yellow and a setting sun red?
Absorbtion and reflection • The wavelength of light reaching our eye from an object is determined by the chemical and physical properties of its surface. • These will determine which wavelengths are absorbed, and which are reflected, and how much each is reflected. • However, while it is generally true that objects which appear red reflect more long wavelength light than any other frequency, this is not always true - and this alone is not what makes it red.
Additive and subtractive colour mixing Additive (eg lights) Subtractive (eg paints)
Hue, intensity and saturation • The phenomenon of colour is more complicated than wavelength judgement anyway. • The wavelength of the light reflected only determines the hue which is seen. • Perceived colour is also determined by the: – intensity of the reflected light (how bright it is) – the saturation of the colour (how much white light is mixed in with the pure hue).
Examples • The difference between blue and red is one of hue. • The difference between light blue and dark blue is (usually) one of intensity. • The difference between red and pink is one of saturation.
A couple of million colours • We (humans) can discriminate roughly 200 hues, 500 intensity steps, and 20 saturations. • When these are combined, that makes about 2 million different colours.
How do we see in colour? • Early theories of colour vision stressed its trichromatic nature. • There is support for this approach, • but it does not explain the most functionally important aspect of colour perception • colour constancy • We will also look at a theory that does account for this - Land’s retinex theory • And a bit at opponent process theory!
Trichromatic theory • (as you already know!) This theory hypothesised that there are three different sorts of receptors and that they respond best to different wavelengths of light. • They respond best to long wavelength (which looks red), medium (green ) or short (blue) wavelength light - this is what cones do! • The colour you see is determined by the relative levels of activity in the three sorts of receptors. • So red objects reflect more long wavelength light than other wavelengths.
Support • This accounts for the results of additive colour mixing experiments R G B
And coloured afterimages
But: Colour constancy • Although it has support - both behavioural and physiological • trichromatic theory completely ignores the most important aspect of colour vision • colours don’t change as levels and balances of illumination change. • This is known as colour constancy. • If colours did change, colour vision would be almost useless
The road to colour constancy: Lightness constancy • If we look at a dark square on a light background in dim illumination (by candlelight, say), then it will look dark. • If we look at the same dark square on the same light background in full sunlight, then it will look just as dark • even though it is now reflecting light which is brighter by a factor of several million.
Why? • Because the background is also reflecting much more light. 100 90 5 10000 9000 500
Two things stay the same • Our perception, • and the ratio of light reflected from the dark vs light regions • this ratio is a environmental constancy, and a perceptual constancy. • We are not sensitive to absolute lightnesses • even though our photoreceptors are.
Simultaneous contrast effect
Colour constancy • Edwin Land discovered that colour vision works the same way. • The clearest demonstration of this is in Mondrian displays The first important result is that the colours don’t change despite wide variations in the levels of red, green and blue illumination
Land’s Mondrian experiments • Find a green patch (say) and record how much green, red and blue light it is reflecting • by using a photometer and by lighting it with one projector at a time • the trichromatic theory would lead us to believe that it is this ratio which determines its greenness • it should be reflecting more green light.
All three projectors on Red only Green only Blue only
The trichromatic theory is wrong! • If we turn our photometer to a blue patch • or a red or yellow or grey one • and adjust the brightness controls on each of the three projectors so that it reflects the same amount of red, green and blue light as the green patch did. . • then when all three projectors are on it should look green. • It doesn’t - it stays blue!
Colour perception is based on a “comparison” of relative lightnesses • As in lightness perception, ratios of lightnesses is what determines perceived colour • in the case of colour it is a “comparison” of the ratios of the lightness as detected by the red, green and blue cone systems. • The perceptual constancy is a ratio of ratios. • So we see a region as red if it reflects more red light than nearby regions (and less blue and green).
All three projectors on Red only Green only Blue only
Opponent Process Theory • This suggested that certain colours were coded as opposites of one another • activation of one half of the pair inhibited activity in the other. • The opposites were red-green, blueyellow and black-white. • We can understand why this might be so in the context of the physiology necessary to implement Land’s theory
Colour processing at the retina • Because they contain different pigments, the three cone types respond best to lights of different wavelengths. • This can be determined physiologically and physically.
Colour processing at the retina • The cones are hooked up to ganglion cells in a fashion which is colour opponent: • R+/G- ganglion cells receive excitatory input from long wavelength cones and an inhibitory connection from medium wavelength cones. • There are R-/G+ cells too. • B+/Y- ganglion cells receive excitatory input from short wavelength cones inhibitory inputs from both long and medium wavelength cones • There are B-/Y+ cells too.
Colour in LGN • Geniculate cells respond just like ganglion cells (as usual), except that the colour opponency is normally spatially delineated • into red-on-centre/green-off-surround, for example. • or yellow-off-centre/blue-on-surround
Colour cells in V 1 • In V 1, there are centre-surround colour opponent cells like in LGN, but there also double colour opponent cells, in which the centre might be R+/G- and the surround R/G+. R+/G- Y+/B- G+/R- B+/Y-
Colour cells in V 1 • The colour sensitive cells in V 1 are clustered into columns of “blobs” which have a particular colour opponency (R/G or B/Y). • The colour sensitive cells, unlike almost any other V 1 cells, don’t care about orientation.
Colour cells in V 4 • There are cells in V 4 which respond in the same way as a human observer in a Land Mondrian experiment. • Their response is not affected by the relative levels of red green and blue lighting the display • They show colour constancy.
Achromatopsia • Damage to V 4 causes achromatopsia - an inability to see any colours
