Texture Analysis and Synthesis

1
Texture Analysis and Synthesis

2
Texture

• Texture pattern thatlooks the sameat all
  locations
• May be structured or random

Wei Levoy
3
Applications of Textures

• Texture analysis
• Detemining statistical properties of textures
• Segmentation
• Recognition
• Shape from texture
• Texture synthesis

4
Oriented Filter Banks
Multiresolution Oriented Filter Bank
OriginalImage
Steerable Pyramid
5
Steerable Pyramid Texture Analysis

• Pass image through filter bank
• Compile histogram of intensitiesoutput by each
  filter
• To synthesize new texture
• Start with random noise image
• Adjust histograms to match original image
• Re-synthesize image from filter outputs

6
Texture Analysis / Synthesis
Original Texture
Synthesized Texture
Heeger and Bergen
7
Textons

• Elements (textons) either identical or come
  from some statistical distribution
• Can analyze in natural images

Olhausen Field
8
Clustering Textons

• Output of bank of n filters can be thought of as
  vector in n-dimensional space
• Can cluster these vectors using k-meansMalik et
  al.
• Result dictionary of most common textures

9
Clustering Textons
Image
Texton to Pixel Mapping
Clustered Textons
Malik et al.
10
Using Texture in Segmentation

• Compute histogram of how many times each of the k
  clusters occurs in a neighborhood
• Define similarity of histograms hi and hj using
  c2
• Different histograms ? separate regions

11
Texture Segmentation
Malik et al.
12
Markov Random Fields

• Different way of thinking about textures
• Premise probability distribution of a
  pixeldepends on values of neighbors
• Probability the same throughout image
• Extension of Markov chains

13
Texture Synthesis Based on MRF

• For each pixel in destination
• Take already-synthesized neighbors
• Find closest match in original texture
• Copy pixel to destination
• Efros Leung 1999,speedup by Wei Levoy

Wei Levoy
14
Shape from Texture

• Look at deformation of individual textons or at
  distribution of textons on a surface
• Perspective distortion strong depth cue
• Foreshortening gives surface normal
• Sparse depth information (only at textons)
• Impose smoothness (regularization) constraint
• Inward/outward ambiguity

15
Shape from Texture Results
Forsyth