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

59
1 Orientation fields and 3D Orientation fields and 3D shape estimation shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics

Transcript of 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological...

Page 1: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

1

Orientation fields and 3D shape Orientation fields and 3D shape estimationestimationRoland W. FlemingMax Planck Institute

for Biological Cybernetics

Page 2: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

2

Cues to 3D ShapeCues to 3D Shape

specularities shading texture

Conventional wisdom: different cues have different physical causes must be processed differently by visual system (‘modules’)

Page 3: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

3

specularities shading texture

Goal: Find commonalities between cues.

Cues to 3D ShapeCues to 3D Shape

Page 4: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

4

Cues to 3D ShapeCues to 3D Shape

Page 5: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

5

Cues to 3D ShapeCues to 3D Shape

Fleming, Torralba, Adelson

Todd and colleagues

Mingolla and Grossberg

Koenderink and van Doorn

Zucker and colleagues

Zaidi and Li

Malik and Rosenholtz

Page 6: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

6

It is remarkable that we can recover 3D shape:

No motion No stereo No shading No texture

image consists of nothing more than a distorted reflection of the world surrounding the object

Ideal mirrored surface

Fleming et al. (2004). JOV

Shape from SpecularitiesShape from Specularities

Page 7: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

7

As the object moves from scene to scene, the image changes dramatically.

Yet, somehow we are able to recover the 3D shape.

Shape from SpecularitiesShape from Specularities

Page 8: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

8

Curvatures determine distortionsCurvatures determine distortions

highly curved

Page 9: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

9

Curvatures determine distortionsCurvatures determine distortions

slightlycurved

Anisotropies in surface curvature lead to powerful distortions of the reflected world

Page 10: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

10

Interpreting distorted reflectionsInterpreting distorted reflections

Page 11: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

11

Orientation fieldsOrientation fields

Ground truth

Page 12: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

12

3D shape appears to be conveyed by the continuously varying patterns of orientation across the image of a surface

Page 13: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

13

Beyond specularityBeyond specularity

Specular reflectionSpecular reflection Diffuse reflectionDiffuse reflection

Page 14: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

14

Differences betweenDifferences betweendiffuse and specular reflectiondiffuse and specular reflection

Page 15: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

15

Differences betweenDifferences betweendiffuse and specular reflectiondiffuse and specular reflection

Page 16: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

16

Differences betweenDifferences betweendiffuse and specular reflectiondiffuse and specular reflection

Page 17: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

17

ShinyShiny

Painted Painted

Page 18: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

18

Beyond specularityBeyond specularity

Specular reflectionSpecular reflection Diffuse reflectionDiffuse reflection

Page 19: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

19

Latent orientationLatent orientationstructurestructure

Page 20: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

20

Orientation fieldsOrientation fieldsin shadingin shading

Page 21: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

21

Orientation fieldsOrientation fieldsin shadingin shading

Page 22: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

22

Reflectance as IlluminationReflectance as Illumination

a(f) = 1 / f

= 0 = 0.4 = 0.8 = 1.2

= 1.6 = 2.0 = 4.0 = 8.0

Page 23: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

23

highly curved

Page 24: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

24

slightlycurved

Anisotropies in surface curvature lead to anisotropies in the image.

Page 25: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

25

Stability across changesStability across changesin surface reflectancein surface reflectance A parametric space of glossy plastic

materials (using Ward model)

Diffuse Reflectance, dDiffuse Reflectance, d

Sp

ecu

lar

Reflect

an

ce,

sS

pecu

lar

Reflect

an

ce,

s

Page 26: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

26

Idea: Experiment 1Idea: Experiment 1

Rationale: measure stability of 3D shape across changes in surface reflectance

Method: gauge figure task? Problem: costly to do full depth reconstruction for

many shapes and materials Solution? Compare sparse gauge measurement? Alternative task?:

locate depth extrema along given raster line (2D task)

Page 27: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

27

TextureTexture

Anisotropic compression of texture depends on surface slant

Page 28: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

28

TextureTexture

Anisotropic compression of texture depends on surface slant

Page 29: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

29

Orientation fieldsOrientation fieldsin texturein texture

Page 30: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

30

Orientation fieldsOrientation fieldsin texturein texture

Page 31: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

31

Orientation fieldsOrientation fieldsin texturein texture

Page 32: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

32

Affine TransformationAffine Transformation

Shear:- does affect first derivatives- does NOT affect second derivatives

Page 33: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

33

Shear:- does affect first derivatives- does NOT affect second derivatives

Affine TransformationAffine Transformation

Page 34: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

34

Shear:- does affect first derivatives- does NOT affect second derivatives

Affine TransformationAffine Transformation

Page 35: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

35

Shear:- does affect first derivatives- does NOT affect second derivatives

Affine TransformationAffine Transformation

Page 36: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

36

Shear:- does affect first derivatives- does NOT affect second derivatives

Affine TransformationAffine Transformation

Page 37: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

37

Shear:- does affect first derivatives- does NOT affect second derivatives

Affine TransformationAffine Transformation

Page 38: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

38

Idea: Experiment 2Idea: Experiment 2

Rationale: use orientation fields to predict misperceptions of 3D shape

Possible methods Gauge figure task?

Matching task: subject adjusts shear of a textured

object until it appears to match the shaded version of the same object

Subject adjusts shear of one oject (shaded or textured) until it appears to match the ‘degree of shear’ of another object? Sounds too strange?

Page 39: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

39

Illusory distortionsIllusory distortionsof shapeof shape

Inspired by Todd & Thaler VSS 05

Page 40: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

40

Illusory distortionsIllusory distortionsof shapeof shape

Inspired by Todd & Thaler VSS 05

Page 41: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

41

Idea: Experiment 3Idea: Experiment 3

Rationale: use orientation fields to predict misperceptions of 3D shape

Possible methods gauge figure task to reconstruct

full 3D shape. Again, this is costly, but perhaps

a few shapes are enough

depth extrema task: locate depth extrema along raster line (this is what Todd and Thaler did). Potentially we could predict the

locus directly from the orientation field

Page 42: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

42

Idea: Experiment 3Idea: Experiment 3 Compare small and large changes in orientation field by using texture stretching along

the line of sight Advantage: same infringement of ‘isotropy assumption’, different change in apparent

3D shape

UnstretchedUnstretchedStretched 2:1Stretched 2:1

along line of sightalong line of sight

Page 43: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

43

Uses biologically plausible measurements

Orientation selectivity maps in primary visual cortex of tree shrew. After Bosking et al. (1997).

Potential of Potential of Orientation FieldsOrientation Fields

Page 44: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

44

No need for visual system to estimate reflectance or illumination explicitly.

Classical shape from shading uses the reflectance map to estimate surface normals from image intensities

Reflectance map is usually unknown and ambiguous

Potential of Potential of Orientation FieldsOrientation Fields

Page 45: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

45

Stable across albedo discontinuities.

Breton and Zucker (1996), Huggins and Zucker (2001)

Potential of Potential of Orientation FieldsOrientation Fields

Page 46: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

46

Handle improbable combinations of reflectance and illumination.

non-linear intensity transfer function

normal shadingnormal shading ‘‘weird’ shadingweird’ shading

Potential of Potential of Orientation FieldsOrientation Fields

Page 47: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

47

We could measures shape estimates with these types of stimuli as well.

non-linear intensity transfer function

normal shadingnormal shading ‘‘weird’ shadingweird’ shading

Link back toLink back toexperiment 1 ?experiment 1 ?

Page 48: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

48

May explain how images with no obvious BRDF interpretation nevertheless yield 3D percepts

Potential of Potential of Orientation FieldsOrientation Fields

Ohad Ben-Shahar

Page 49: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

49

Converting between cuesConverting between cues

input imageinput image

Todd & Oomes 2004

( )2

Latent shadingLatent shading

Page 50: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

50

( )2

Converting between cuesConverting between cues

input imageinput image

Todd & Oomes 2004

Latent shadingLatent shading

Page 51: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

51

ConclusionsConclusions

Orientation fields are potentially a very powerful source of information about 3D shape

For the early stages of 3D shape processing, seemingly different cues may have more in common than previously thought

Page 52: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

52

Thank youThank youCollaborators

Ted AdelsonAntonio Torralba

Funding

RF supported byDFG FL 624/1-1

Page 53: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

53

What still needsWhat still needsto be explained?to be explained?

For Lambertian materials (or blurry illuminations), the reflectance map is so smooth that it is significantly anisotropic.

Therefore shading orientation fields vary considerably with changes in illumination.

sidefront top

Page 54: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

54

What still needsWhat still needsto be explained?to be explained?

Surprising prediction: 3D shape should actually be less stable across changes in illumination for diffuse than for specular materials.

We found evidence for changes in 3D shape with changes in illumination Alternative: higher order invariants establish an equivalence between

different orientation fields. Example: joint measures of orientation at different locations.

sidefront top

Page 55: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

55

Note analogy to textures of different orientations

Todd et al. (2004)

What still needsWhat still needsto be explained?to be explained?

Page 56: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

56

Matte dark grey

Rough metal

Glossy light grey

Page 57: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

57

PlasticsPlastics

(a) Mirror (b) Smooth plastic (c) Rough plastic

Page 58: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

58

When the world is anisotropicWhen the world is anisotropic

Brushed horizontally Brushed vertically

Page 59: 1 Orientation fields and 3D shape estimation Roland W. Fleming Max Planck Institute for Biological Cybernetics.

59

Stability across changesStability across changesin surface reflectancein surface reflectance A parametric space of glossy plastic

materials (using Ward model)

Diffuse Reflectance, dDiffuse Reflectance, d

Sp

ecu

lar

Reflect

an

ce,

sS

pecu

lar

Reflect

an

ce,

s