Contourlet Transforms For Feature Detection

1
Contourlet Transforms For Feature Detection

• Wei-shi Tsai
• April 29th, 2008

2
Feature Detection

• Focus will be on edge detection
• Gradient operators (Sobel, Roberts)
• Laplacian operators
• LoG (Laplacian of Gaussian)
• DoG (Difference of Gaussians)
• Canny method
• Anisotropic diffusion

3
Contourlets (Do and Vetterli, 2005)

• Captures smooth contours and edges at any
  orientation
• Filters noise
• Derived directly from discrete domain instead of
  extending from continuous domain
• Can be implemented using filter banks

4
Contourlet filter bank
The transform decouples the multiscale and the
directional decompositions.
5
Test Pattern Image Scale 1
6
Test Pattern Image Scale 2
7
Peppers Image Scale 1
8
Peppers Image Scale 2
9
Generic Girl Image Scale 1
10
Generic Girl Image Scale 2
11
Tiffany Image Scale 1
12
Tiffany Image Scale 2
13
Elaine Image Scale 1
14
Elaine Image Scale 2
15
Lena Image Scale 1
16
Lena Image Scale 2
17
Conclusions

• Contourlet transforms can be used for edge
  detection
• Results can vary based on the type of image
• Evaluation is only useful given what the feature
  extracted is to be used