Perception of illumination and shadows Lavanya Sharan February 14th, 2011.

Slides:



Advertisements
Similar presentations
1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.
Advertisements

Lecture 16 – Lightness Perception
Visual Perception of 3D Shape Roland W. Fleming Manish Singh Max Planck Institute for Biological Cybernetics Rutgers University – New Brunswick.
Announcements Project 2 due today Project 3 out today –demo session at the end of class.
Computer Vision Spring ,-685 Instructor: S. Narasimhan Wean 5403 T-R 3:00pm – 4:20pm Lecture #12.
3D Modeling from a photograph
3D Graphics Rendering and Terrain Modeling
SPECULAR FLOW AND THE PERCEPTION OF SURFACE REFLECTANCE Stefan Roth * Fulvio Domini † Michael J. Black * * Computer Science † Cognitive and Linguistic.
Shape-from-X Class 11 Some slides from Shree Nayar. others.
Outline Sensation, Perception, Behavior Process of sensation Perceived vs. “real” world Properties of perceptual processes - Adaptation, pattern coding.
PSY 5018H: Math Models Hum Behavior, Prof. Paul Schrater, Spring 2004 Vision as Optimal Inference The problem of visual processing can be thought of as.
Recovering Intrinsic Images from a Single Image 28/12/05 Dagan Aviv Shadows Removal Seminar.
May 2004SFS1 Shape from shading Surface brightness and Surface Orientation --> Reflectance map READING: Nalwa Chapter 5. BKP Horn, Chapter 10.
Shadows (P) Lavanya Sharan February 21st, Anomalously lit objects are easy to see Kleffner & Ramchandran 1992Enns & Rensink 1990.
Visibility Subspaces: Uncalibrated Photometric Stereo with Shadows Kalyan Sunkavalli, Harvard University Joint work with Todd Zickler and Hanspeter Pfister.
Features and Object in Visual Processing. The Waterfall Illusion.
What is color for?.
Perception of illumination and shadows Lavanya Sharan February 14th, 2011.
Color, lightness & brightness Lavanya Sharan February 7, 2011.
Materials II Lavanya Sharan March 2nd, Computational thinking about materials Physics-basedPseudo physics-based.
Features and Object in Visual Processing. The Waterfall Illusion.
Texture perception Lavanya Sharan February 23rd, 2011.
CSCE 641 Computer Graphics: Radiosity Jinxiang Chai.
Photometric Stereo & Shape from Shading
Image Statistics and the Perception of 3D Shape Roland W. Fleming Max Planck Institute for Biological Cybernetics Yuanzhen Li Edward H. Adelson Massachusetts.
Project 4 Results Representation – SIFT and HoG are popular and successful. Data – Hugely varying results from hard mining. Learning – Non-linear classifier.
1 Perceiving 3D from 2D Images How can we derive 3D information from one or more 2D images? There have been 2 approaches: 1. intrinsic images: a 2D representation.
Computer Vision Spring ,-685 Instructor: S. Narasimhan PH A18B T-R 10:30am – 11:50am Lecture #13.
CAP4730: Computational Structures in Computer Graphics 3D Concepts.
Technology and Historical Overview. Introduction to 3d Computer Graphics  3D computer graphics is the science, study, and method of projecting a mathematical.
Computer Graphics Psychophysics Heinrich H. Bülthoff Max-Planck-Institute for Biological Cybernetics Tübingen, Germany Heinrich H. Bülthoff Max-Planck-Institute.
Analysis of Lighting Effects Outline: The problem Lighting models Shape from shading Photometric stereo Harmonic analysis of lighting.
1 Perception and VR MONT 104S, Fall 2008 Session 13 Visual Attention.
Lecture 2b Readings: Kandell Schwartz et al Ch 27 Wolfe et al Chs 3 and 4.
Radiance Workshop, October 1-2, 2007 Perceived shininess and rigidity - Measurements of shape-dependent specular flow of rotating objects Katja Doerschner.
Image Filtering Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem 02/02/10.
Goal and Motivation To study our (in)ability to detect inconsistencies in the illumination of objects in images Invited Talk! – Hany Farid: Photo Forensincs:
PERCEPTUAL STRATEGIES FOR MATERIAL IDENTIFICATION Qasim Zaidi Rocco Robilotto Byung-Geun Khang SUNY College of Optometry.
Color and Brightness Constancy Jim Rehg CS 4495/7495 Computer Vision Lecture 25 & 26 Wed Oct 18, 2002.
Perceptual Constancy Module 19. Perceptual Constancy Perceiving objects as stable or constant –having consistent lightness, color, shape, and size even.
Shade & Shadow Figure 1-3, Page 11
Cornell CS465 Spring 2004 Lecture 4© 2004 Steve Marschner 1 Shading CS 465 Lecture 4.
1Ellen L. Walker 3D Vision Why? The world is 3D Not all useful information is readily available in 2D Why so hard? “Inverse problem”: one image = many.
Cognitive - perception.ppt © 2001 Laura Snodgrass, Ph.D.1 Perception The final image we are consciously aware of is a “constructed” representation of the.
To Students in Psychology Some of you have used part of at least one Trinity Day learning about color perception on your own. The internet is.
From local motion estimates to global ones - physiology:
MAN-522 Computer Vision Spring
Visual computation of lightness in simple and complex images
Instructor: S. Narasimhan
SPECULAR FLOW AND THE PERCEPTION OF SURFACE REFLECTANCE
Prof. Riyadh Al_Azzawi F.R.C.Psych
3D Graphics Rendering PPT By Ricardo Veguilla.
Merle Norman Cosmetics, Los Angeles
Common Classification Tasks
Image formation and the shape from shading problem
Image formation and the shape from shading problem
Questions for lesson 3 Perception 11/27/2018 lesson 3.
L Differentiate Threshold Integrate R.
A preference for global convexity in local shape perception
Prof. Riyadh Al_Azzawi F.R.C.Psych
Algorithm and perception
Physical Properties of light
Depthmap Reconstruction Based on Monocular cues
Visual Motion and the Perception of Surface Material
What fraction of the incident light is reflected toward the viewer?
Announcements Project 3 out today demo session at the end of class.
Analysis of Complex Designs
CSE (c) S. Tanimoto, 2007 Image Understanding
Prof. Riyadh Al_Azzawi F.R.C.Psych
Shape from Shading and Texture
Presentation transcript:

Perception of illumination and shadows Lavanya Sharan February 14th, 2011

Studied indirectly Not a lot of studies examine illumination or shading directly Role of illumination and shading in perception of 3-D shape, reflectance, object identity and space

Outline Shape from shading Illumination estimation Shadows

Shape from shading is under- constrained. Fig 9.11, VPfaCGP And yet, we perceive unique and stable shapes.

Theoretical cues for shape from shading Reflectance map (Horn, 1977) Isophotes (Koenderink & Van Doorn, 1980) Image orientation (Fleming, Torralba & Adelson, 2004)

Theoretical cues for shape from shading Reflectance map (Horn, 1977) Representation of scene brightness as a function of 3-D surface orientation Ignores shadows, inter-reflections, vignetting, translucency etc. Unclear whether this relationship between image intensity & surface orientation is used by visual system Horn & Sjoberg, 1978

Theoretical cues for shape from shading Isophotes (Koenderink & Van Doorn, 1980) Curves of constant intensity, depend on illumination and shape Patterns of isophotes can reveal shape (under assumptions of lighting) The visual system could use these as a cue Fig 9.12, VPfaCGP

Theoretical cues for shape from shading Image orientation (Fleming, Torralba & Adelson, 2004) Orientation filters have strong responses for strong curvature regions. By measuring these across a surface can get local geometry The visual system could use this relationship between image orientation and surface curvature. Fig 9.13, VPfaCGP

Why is shape from shading hard? Lots of ambiguities. Convex vs. concave? Surface orientation change vs. surface reflectance change? Bas-relief ambiguity

Ambiguities in shape-from-shading Convex vs. concave? (Ramachandran, 1988)

Reflectance vs. orientation change? (Knill & Kersten, 1991) Ambiguities in shape-from-shading

Bas-relief ambiguity (Belhumeur et al., 1999) Ambiguities in shape-from-shading

Gauge figure task to study shape perception (Koenderink et al., 1992)

What have we learnt from gauge figure tasks? Subjects are consistent. Their (inferred) shapes are related by affine transforms. (Koenderink et al., 1992) For simple shapes, contours are often enough for estimating shape, shading plays a lesser role. (Mamassian & Kersten., 1996; Koenderink et al., 1996; Cole et al., 2009) Illumination changes causes subtle distortions of perceived shape. (Koenderink et al., 1996; Caniard & Fleming, 2007)

Intrinsic image analysis Fig 9.15, VPfaCGP Idea: Visual system separates retinal image into layers that represent distinct physical causes. (Barrow & Tenenbaum, 1978) How? Proposals include Retinex, anchoring theory, etc.

Mutual illumination affects reflectance perception Ruppertsberg & Bloj, 2007 We can distinguish black and white rooms seen in isolation based on inter- reflections. (Gilchrist & Jacobsen, 1984) Mutual illumination estimation is not perfect, sometimes perceived as surface color. (Bloj et al., 1999; Doerschner et al., 2004)

Testing illumination perception directly Koenderink et al., 2003 Subjects can tell direction (upto convex/concave ambiguity. Worse at estimating elevation of light source (related to bas-relief ambiguity) Confirmed by other studies (Ho et al., 2006; Pont & Koenderink 2003)

Testing illumination perception directly Subjects can estimate diffuseness of a light source, errors related to collimated/diffuse. (Pont & Koenderink, 2007) Can account for two collimated light sources (Doerschner et al., 2007) Can to some extent account for changes across space (Snyder et al., 2005) and time (Gerhard & Maloney, 2010)

Light from above prior (Ramachandran, 1988) A is usually seen as convex => light from above Discs in B cannot be both convex or concave => one light source

Light from above and left prior Time to search in display depended on the shading direction of distractors. Should be lowest when shading matches prior. (Sun & Perona, 1998)

Shadows help interpret scenes Kersten et al., 1999 The spatial position of objects is influenced by cast shadows

Shadows help interpret scenes (Let’s watch a movie )

Last word about shadows Do these results hold for real-world images?

Surprisingly hard to see errors (Ostrovsky et al., 2005)

(Lopez-Moreno et al., 2010) Need to understand this better...

Summary ✓ We can estimate illumination, tested indirectly by probing reflectance and shape, and also directly. ✓ We are not perfect, but the problem is also hard (lots of ambiguities). We are consistent in our estimates, e.g., light-from-above(-and-left) prior. ✓ Shading tells us about shape, shadows about scene. ✓ Don’t know a lot about cues and effects in real- world scenes.