EECE253_04_ColorPerception

EECE253_04_ColorPerception - EECE/CS 253 Image Processing...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: EECE/CS 253 Image Processing Lecture Notes: Lecture Notes on Color Perception Richard Richard Alan Peters II Peters II Department of Department of Electrical Engineering and Engineering and Computer Science Computer Science Fall Fall Semester 2011 2011 . . . . . . . . . This work is licensed under the Creative Commons Attribution-Noncommercial 2.5 License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc/2.5/ or send a letter to Creative Commons, 543 Howard Street, 5th Floor, San Francisco, California, 94105, USA. Color Images Are constructed from three intensity maps. Each intensity map is projected through a color filter (e.g., red, green, or blue, or cyan, magenta, or yellow) to create a single color image. The intensity maps are overlaid to create a color image. Each pixel in a color image is a three element vector. 2011-09-14 1999-2011 by Richard Alan Peters II 2 Color Images on a CRT or LCD Display Intensity images are projected through dot-array color filters which are slightly offset from one another. Projected image primary colors: red, green, and blue. 2011-09-14 1999-2011 by Richard Alan Peters II 3 Color Images on a CRT or LCD Display Photographs of various displays, showing various pixel geometries. Clockwise from top left, a standard definition CRT television, a CRT computer monitor, a laptop LCD, and the OLPC XO-1 LCD display. [Peter Halasz (user:Pengo), Wikipedia, http://en.wikipedia.org/wiki/Pixel_geometry] 2011-09-14 1999-2011 by Richard Alan Peters II 4 Color Images In Print Images are separated into four color bands, each of which is printed as a grid regularly spaced dots. A dot’s diameter varies in proportion to the intensity of the color. 2011-09-14 1999-2011 by Richard Alan Peters II 5 Color Images in Print The four colors are magenta, cyan, yellow, and black 2011-09-14 1999-2011 by Richard Alan Peters II 6 Standard Halftone Screen Angles The dot grids are created with a screen that overlays the intensity images. Cyan: 105° Yellow: 90° Magenta: 75° Black: 45° The screens are oriented at different angles. The resulting patterns are called “rosettes”. 2011-09-14 1999-2011 by Richard Alan Peters II 7 Color Separation / Halftoning The original is separated into an intensity image for each of the four color bands. 2011-09-14 1999-2011 by Richard Alan Peters II 8 Color Separation / Halftoning 2011-09-14 1999-2011 by Richard Alan Peters II 9 Color Separation / Halftoning Each intensity image is multiplied by a corresponding “screen”, Cyan Magenta Yellow Black Each screened image is printed in its own color on the same page. 2011-09-14 1999-2011 by Richard Alan Peters II 10 Color Separation / Halftoning 2011-09-14 1999-2011 by Richard Alan Peters II 11 The Eye Diagram from http://webvision.med.utah.edu/ 2011-09-14 1999-2011 by Richard Alan Peters II 12 The Retina Fovea Macula Optic nerve Diagram from http://webvision.med.utah.edu/ 2011-09-14 1999-2011 by Richard Alan Peters II 13 The Retina Light Diagram from http://webvision.med.utah.edu/ 2011-09-14 1999-2011 by Richard Alan Peters II 14 Retinal Mosaic Cepko, Connie, “Giving in to the blues”, Nature Genetics, 24, 99 - 100 (2000) cepko@genetics.med.harvard.edu 2011-09-14 1999-2011 by Richard Alan Peters II 15 Photoreceptor Densities Diagrams from http://webvision.med.utah.edu/ 2011-09-14 1999-2011 by Richard Alan Peters II 16 Photoreceptor Densities The density of cone photoreceptors decreases from the high-resolution fovea to the periphery of the eye. A human eye’s field of view is about 155° of that, the fovea comprises the central 2°. To see the world in detail requires active scanning by the eyes. A person does not see much more than he or she does see in most situations. The slides that follow mimic a multiresolution scan of a painting by a single eye. (The digital image processing in this case was done with a log-polar transform.) Figure: Anatomical Distribution of Rods and Cones from Neuroscience. 2nd edition. Purves D, Augustine GJ, Fitzpatrick D, et al., editors. Sunderland (MA): Sinauer Associates; 2001. http://www.ncbi.nlm.nih.gov/books/NBK10848/ 2011-09-14 1999-2011 by Richard Alan Peters II 17 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 18 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 19 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 20 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 21 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 22 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 23 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 24 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Log-polar transform applied. 25 1999-2011 by Richard Alan Peters II 2011-09-14 Retinal Space-Variant Sensing Louis Boilly (1761-1845) Thirty-Six Faces of Expression. 2011-09-14 1999-2011 by Richard Alan Peters II 26 Retina: CenterSurround Edge Detector spatial domain ∇ g (r ) = 2 The interconnection of the photoreceptors by the other cells in the retina cause its output to be an edge map, similar to the action of a Laplacian of Gaussian filter on a digital image. 2011-09-14 σ = 32 1 πσ 4 frequency domain r2 r2 1− exp − 2σ 2 2σ 2 σ=2 G (ω ) = ω exp − 2 σ 2ω 2 2 Laplacian of Gaussian (LoG) Filter r2 = x2 + y2 1999-2011 by Richard Alan Peters II ω 2 = u2 + v2 27 Retinal Edge Detection Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Photo negative of LoG output. 28 1999-2011 by Richard Alan Peters II 2011-09-14 Space Variant Retinal Edge Detection Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Photo negative of LoG output. 29 1999-2011 by Richard Alan Peters II 2011-09-14 The Retinal Transform Minimizes Data Bandwidth Louis Boilly (1761-1845) Thirty-Six Faces of Expression. Photo negative of LoG output. 2011-09-14 This is the reduction in size from the full image to a compact multiresolution representation including the fovea (the disk) and the periphery. 1999-2011 by Richard Alan Peters II 30 L – downsample factor R – information content Pixelization of Color Images: All Bands Equal 2011-09-14 1999-2011 by Richard Alan Peters II Photo: R. A. Peters II, 1998, The Lake, Central Park, NYC. 31 L – downsampled band R – information content 16× Pixelization of Color Images: R, G, & B Bands 2011-09-14 1999-2011 by Richard Alan Peters II Photo: R. A. Peters II, 1998, The Lake, Central Park, NYC. 32 Visual Areas in the Brain Retina: center-surround color feature detectors LGN: (lateral geniculate nucleus) relay to V1; audio attention V1: selective spatiotemporal filters V2: feature aggregation V3: visual attention IT: (Inferior temporal gyrus) complex object features Graphic from M. Lewicky 2011-09-14 1999-2011 by Richard Alan Peters II 33 In the Brain: from RGB to LHS luminance The eye has 3 types of photoreceptors: sensitive to red, green, or blue light. hue saturation The brain transforms RGB into separate brightness and color channels (e.g., LHS). brain 2011-09-14 photo receptors 1999-2011 by Richard Alan Peters II 34 L – downsample factor R – information content 16× Pixelization of Color Images: Luminance Only 2011-09-14 1999-2011 by Richard Alan Peters II Photo: R. A. Peters II, 1998, The Lake, Central Park, NYC. 35 L – downsample factor R – information content 16× Pixelization of Color Images: Chrominance (H+S) Only 2011-09-14 1999-2011 by Richard Alan Peters II Photo: R. A. Peters II, 1998, The Lake, Central Park, NYC. 36 16× Pixelization L – downsampled band R – information content 2011-09-14 1999-2011 by Richard Alan Peters II Photo: R. A. Peters II, 1998, The Lake, Central Park, NYC. 37 16× Pixelization L – downsampled band R – information content 2011-09-14 These 4 images all have the same amount of digital information… 1999-2011 by Richard Alan Peters II Photo: R. A. Peters II, 1998, The Lake, Central Park, NYC. 38 16× Pixelization L – downsampled band R – information content 2011-09-14 … but different visual information. 1999-2011 by Richard Alan Peters II Photo: R. A. Peters II, 1998, The Lake, Central Park, NYC. 39 Color Sensing / Color Perception These are approximations of the responses to the visible spectrum of the “red”, “green”, and “blue” receptors of a typical human eye. 2011-09-14 1999-2011 by Richard Alan Peters II 40 Color Sensing / Color Perception Note that the “red” receptor exhibits the same response at 4 different wavelengths … 2011-09-14 1 2 3 1999-2011 by Richard Alan Peters II 4 41 Color Sensing / Color Perception … but the responses of the “green” and “blue” receptors differ … 1 2 3 4 b1 r1 g1 2011-09-14 1999-2011 by Richard Alan Peters II 42 Color Sensing / Color Perception … at each of the 4 locations so that … 1 2 3 4 b2 r2 g2 2011-09-14 1999-2011 by Richard Alan Peters II 43 Color Sensing / Color Perception … each of the 4 wavelengths is represented by a unique response from the set of 3 receptors. 1 2 3 4 g3 r3 b3 2011-09-14 1999-2011 by Richard Alan Peters II 44 Color Sensing / Color Perception 1 2 3 4 r4 g4 b4 2011-09-14 1999-2011 by Richard Alan Peters II 45 Color Sensing / Color Perception These are approximations of the responses to the visible spectrum of the “red”, “green”, and “blue” receptors of a typical human eye. The simultaneous red + blue response causes us to perceive a continuous range of hues on a circle. No hue is greater than or less than any other hue. 2011-09-14 1999-2011 by Richard Alan Peters II 46 Complementary Colors CYAN - RED GREEN - MAGENTA BLUE - YELLOW Colors opposite each other on the color disk are called “complementary”. 2011-09-14 1999-2011 by Richard Alan Peters II 47 Complementary Colors red green blue response color 0° color 180° red green blue response photoreceptor response is represented as proportional to brightness To complementary colors, the response of the retina’s photoreceptors is opposite. 2011-09-14 1999-2011 by Richard Alan Peters II 48 Color Perception: The Afterimage Effect Stare at the dot in the center of the image 2011-09-14 1999-2011 by Richard Alan Peters II 49 2011-09-14 1999-2011 by Richard Alan Peters II 50 2011-09-14 1999-2011 by Richard Alan Peters II 51 2011-09-14 1999-2011 by Richard Alan Peters II 52 Color Perception: The Afterimage Effect The color “negatives” saturate the local receptors so that when the color is removed the agonist (opposite) color receptors remain saturated. 2011-09-14 1999-2011 by Richard Alan Peters II 53 Color Perception: the Cornsweet Effect Dale Purves, R. Beau Lotto, Surajit Nundy, “Why We See What We Do”, American Scientist, Volume 90, No. 3, May-June 2002 2011-09-14 1999-2011 by Richard Alan Peters II 54 Color Perception: the Cornsweet Effect The top is darker… …than the bottom, Right? Dale Purves, R. Beau Lotto, Surajit Nundy, “Why We See What We Do”, American Scientist, Volume 90, No. 3, May-June 2002 2011-09-14 1999-2011 by Richard Alan Peters II 55 Color Perception: the Cornsweet Effect Wrong! Dale Purves, R. Beau Lotto, Surajit Nundy, “Why We See What We Do”, American Scientist, Volume 90, No. 3, May-June 2002 2011-09-14 1999-2011 by Richard Alan Peters II 56 Color Perception: the Cornsweet Effect Dale Purves, R. Beau Lotto, Surajit Nundy, “Why We See What We Do”, American Scientist, Volume 90, No. 3, May-June 2002 2011-09-14 1999-2011 by Richard Alan Peters II 57 Brightness Perception 255 0 image intensity profile Linear intensity changes are not seen as such. 2011-09-14 1999-2011 by Richard Alan Peters II 58 Brightness Perception The previous slide demonstrates the WeberFechner relation. The linear slope of the intensity change is perceived as logarithmic. g1 - g 2 Dg = g1 + g 2 The green curve is the actual intensity; the blue curve is the perceived intensity. 2011-09-14 1999-2011 by Richard Alan Peters II 59 decreasing contrast Uniform Change in Frequency and Contrast increasing frequency 2011-09-14 1999-2011 by Richard Alan Peters II 60 ...
View Full Document

Ask a homework question - tutors are online