Outline • Color perception – Introduction – Theories of color perception • Yong-Helmhotz trichromatic theory • Hering opponent process theory • Dual process theory – Physiological mechanisms for color perception Color Perception • The sensation of color is caused by the brain. • One way to get it is the response of the eye to the presence/absence of light at various wavelengths. 5/29/2016 Visual Perception Modeling 2 Physical description of light • Photon – A tiny packet of vibrating electro-magnetic energy – Characterized by the wavelength of its vibration – The photons we experience in visible light cover just a small portion of the electro-magnetic spectrum 5/29/2016 Visual Perception Modeling 3 Light Spectra 5/29/2016 Visual Perception Modeling 4 Physical description of light • Photon – A tiny packet of vibrating electro-magnetic energy – Characterized by the wavelength of its vibration – The photons we experience in visible light cover just a small portion of the electro-magnetic spectrum • Physical description of light – Number of photons it contains at each wavelength 5/29/2016 Visual Perception Modeling 5 Sunlight 5/29/2016 Visual Perception Modeling 6 Psychological description of color • Color space – The subjective experience of surface colors can be described in terms of three dimensions • Hue • Saturation • Lightness – Color space is the three-dimensional coordinate system in which each color experience is represented by a point 5/29/2016 Visual Perception Modeling 7 Psychophysical correspondence • Mapping between physical descriptions and psychological ones – Mean wavelength determines hue – Spectral area determines lightness • The total number of photons – Variance determines saturation 5/29/2016 Visual Perception Modeling 8 Basic Phenomena • Light mixture – Only a small portion of the colors correspond to monochromatic lights – Two or more colors must be combined in order to non-spectral colors and de-saturated colors – By experience, it is possible to match almost all colors using only three primary sources 5/29/2016 Visual Perception Modeling 9 Additive and subtractive matching • Choose colors A, B, C such that no two can be mixed to match the third - Primaries. • Many colors can be represented as a mixture of A, B, C write M=a A + b B + c C • This is additive matching. • Gives a color description system - two people who agree on A, B, C need only supply (a, b, c) to describe a color. 5/29/2016 • Some colors can’t be matched like this: instead, must write M+a A = b B+c C • This is subtractive matching. • Interpret this as (-a, b, c) • Problem for building monitors: Choose R, G, B such that positive linear combinations match a large set of colors Visual Perception Modeling 10 Some Phenomena • Color blindness – People who cannot discriminate among all the colors – There are several distinct varieties of color blindness • Color afterimages • Mach bands • Chromatic adaptation 5/29/2016 Visual Perception Modeling 11 Early theories and confusion • Widespread: We see things by rays fired out of the eyes, hitting surfaces; colors come from different kinds of rays. • Newton: Color is obtained by differential refraction of white light. • Artists: Can’t get enough colors from Newton; actually, color is obtained by mixing lights or pigments. 5/29/2016 Visual Perception Modeling 12 Young-Helmholtz Trichromatic Theory • Propose by Young-Helmholtz – There are three types of color receptors in the human eye – They were hypothesized to produce the psychologically primary color sensations of red, green, and blue – All other colors were explained as combinations of these primaries 5/29/2016 Visual Perception Modeling 13 Opponent Process Theory • Trichromatic theory cannot explain very well some facts and observations – Yellow is the additive mixture of red and green primaries but the subjective experience does not appear to be that way – Color experiences are always lost in certain pairs 5/29/2016 Visual Perception Modeling 14 Opponent Process Theory – cont. • Opponent process theory by Hering – There are four chromatic primaries rather than three – They are structured in pairs of polar opposites • Red/green • Blue/yellow • Black/write 5/29/2016 Visual Perception Modeling 15 Dual Process Theory • For many decades there were heated debates between two warring factions – Helmholtz vs. Hering • In 1957, Hurvich and Jameson proposed a dual processing theory Trichromatic theory 5/29/2016 Opponent-Process stage Visual Perception Modeling 16 Physiological Mechanisms • Three cone systems – There are three types of cones in the normal trichromat retina, each of which contains a different light-absorbing pigment – Absorption spectra of these three cone types are determined using different techniques 5/29/2016 Visual Perception Modeling 17 Color Receptors “Red” cone 5/29/2016 “Green” cone Visual Perception Modeling “Blue” cone 18 Color Opponent Cells • Color opponent cells in LGN – Responses in the LGN of moneys are roughly conformed to the pattern predicted by Hering’s opponent process theory – They are also found in the bipolar and ganglion cells in the retina 5/29/2016 Visual Perception Modeling 19 Theory for Color Perception • Trichromatic theory is perhaps the only theory that explains many perceptual phenomena nicely and in a unified framework • People have been trying to find theories in the other areas of perception – Texture perception 5/29/2016 Visual Perception Modeling 20 Representing Colors • Accurate color reproduction is commercially valuable - e.g. Kodak yellow, painting a house. • Of the order of 10 color names are widely recognized by English speakers - other languages have fewer/more, but not much more. • Color reproduction problems increased by prevalence of digital imaging – e.g. digital libraries of art. • Choosing pixel values to reproduce/evoke experiences, e.g. an architectural model. • Consistency in user interfaces, monitor-printer consistency, etc. 5/29/2016 Visual Perception Modeling 21 Color spaces • Linear color spaces describe colors as linear combinations of primaries – Choice of primaries • choice of color matching functions • choice of color space • RGB – primaries are monochromatic, energies are 645.2nm, 526.3nm, 444.4nm – Color matching functions have negative parts -> some colors can be matched only subtractively. • CIE XYZ – Color matching functions are positive everywhere, but primaries are imaginary 5/29/2016 Visual Perception Modeling 22 CIE x, y 5/29/2016 Visual Perception Modeling 23 Qualitative features of CIE x, y • Linearity implies that colors obtainable by mixing lights with colors A, B lie on line segment with endpoints at A and B • Monochromatic colors (spectral colors) run along the “Spectral Locus” • Dominant wavelength = Spectral color that can be mixed with white to match 5/29/2016 • Purity = (distance from C to spectral locus)/(distance from white to spectral locus) • Wavelength and purity can be used to specify color. • Complementary colors=colors that can be mixed with C to get white Visual Perception Modeling 24 More linear color spaces • Monitor RGB: primaries are monitor phosphor colors, primaries and color matching functions vary from monitor to monitor - careful! • YIQ: mainly used in television, Y is (approximately) intensity, I, Q are chromatic properties. Linear color space; hence there is a matrix M that transforms XYZ coords to YIQ coords. I and Q can be transmitted with low bandwidth. 5/29/2016 Visual Perception Modeling 25 Non-linear color spaces • HSV: Hue, Saturation, Value are non-linear functions of XYZ. – because hue relations are naturally expressed in a circle • Uniform: equal (small!) steps give the same perceived color changes. • Munsell: describes surfaces, rather than lights - less relevant for graphics. Surfaces must be viewed under fixed comparison light 5/29/2016 Visual Perception Modeling 26 Color books 5/29/2016 Visual Perception Modeling 27 Device independent color imaging • Problem: ensure that colors on a display, printer, etc. give the same experience that a viewer would have seeing relevant light spectra • Difficulty: limited gamuts of most output devices • Strategy: exploit a model of human experience 5/29/2016 Visual Perception Modeling 28 Constancy • We observe the color of • Lightness constancy the light reflected from – how light is the surface, independent of the surfaces brightness of the • But we want surface illuminant colour – problem is known as colour constancy – issues • spatial variation in illumination • absolute standard – Human lightness constancy is very good 5/29/2016 Visual Perception Modeling 29 Finding Skin Using Image Color • It is very useful to find human skin in images – Gesture-based user interfaces – Content-based retrieval – Ignore some pictures • A computational model of skin 5/29/2016 Visual Perception Modeling 30 Surface Color from Image Color • The color of light in an image is determined by two factors – Spectral reflectance of the surface – Spectral radiance of the light falling on that surface • Color of light falling surfaces can vary very widely – Image color can be a bad representation of the color of surfaces being viewed • Color constancy in human perception 5/29/2016 Visual Perception Modeling 31