Texture Recognition and Synthesis

1
Texture Recognition and Synthesis

• A Non-parametric Multi-Scale Statistical Model
• by
• De Bonet Viola
• Artificial Intelligence Lab
• MIT
• Presentation by Pooja

2
Main Goal

• Train on example images
• Recognize novel images
• Generate new images
• How?

3
Markov Random Fields (MRFs)

• Based on simple, local interactions
• Success in restoration
• Weak generative properties
• Inability to capture long range interactions

4
Wavelet Transform

• Effective for modeling natural images
• Measures the underlying causes of images
• Assumption causes are statistically independent
• Coefficients are uncorrelated

5
Multi-scale Wavelet Techniques

• Iterative convolution of bank of filters
• Pyramid of low frequency downsampled images
• Images are a linear transform of statistically
  independent causes

6
Texture Synthesis

• Bergen and Heeger
• Inverse wavelet transform

7
Other synthesis failures
Find me!
8
Other synthesis failures (contd)
Find me!
9
Other synthesis failures (contd)
Hehehe, find me this time!!
10
Think Think Think!

• Why/When does synthesis fail?
• What does it tell us about requirements in a
  successful synthesis technique?

11
Objective of Synthesis

• Different from the original
• Generated by the same underlying stochastic
  process

12
Back to Texture Recognition

• Gauravll do all the synthesis explaining (Thank
  God!)

13
Wavelet coefficients not independent

• Long edges?
• Parent vector of a pixel defined as

14
Probabilistic model

• Generation of nearby pixels strongly dependent

15
Conditional Distributions

• Estimated as a ratio of Parzen window density
  estimators

16
Cross Entropy (Motivation)

• Biased coin p(h) 0.75
• 1st output h 75, t 25
• 2nd output h 100, t 0
• Which is more likely?
• Which is more typical?
• Concept of cross entropy (Kullback-Liebler
  divergence)

17
Cross Entropy

• Viewed as the difference between two expected log
  likelihoods
• Replace integral with monte-carlo sampling
• Lowest cross entropy vs. false positives (or
  negatives)?

18
Results

• Standardized tests
• easy data sets Brodatz texture test suite
• 100 correct classification
• Natural textures
• 20 types of natural texture
• 87 correct classification (humans 93)

19
Pros Cons

• Pros
• Pyramidal dependency
• Concept of likely vs. typical
• False positives vs. overall low cross entropy
• Cons
• Estimation of conditional distributions?
• The R(.) function?
• Works well on simple texture sets! So?
• Only 20 natural textures??
• Errors in the paper ?

20
And finally.

• A not so boring slide!
• Do you think their method would work on this
  texture??