Image Features, Hough Transform

1

• Image Features, Hough Transform
• Image Pyramid
• CSE399b, Spring 06
• Computer Vision
• Lecture 10

http//www.quicktopic.com/35/H/NHVD8SZQQJHZ
2
Boundary and Edge Edge detection-gt lines
3
An example
S.F. in fog
S.F. in Canny
4
An example
S.F. in fog
S.F. with Hough lines
5
Hough Transform

• image edges needs to be grouped into lines and
  junctions
• Hough transform Detect lines in an edge image

6
Line Representation

• is the distance from the origin to the line
• is the norm direction of the line
• Image space Hough space
• point in image space gt a curve in hough space

7
Line Representation

• is the distance from the origin to the line
• is the norm direction of the line
• Image space Hough space
• point in image space gt a curve in hough space

For every theta, set
8
Hough Space

• point in hough space gt line in image space

9
Intersection of the curves

• Each pixel in the image gt
• One curve in Hough space
• What is the intersection of the curves?

10
Hough Transform

• Points in the line
• In hough space, all the curves pass
• So the intersection of the curves is the
  parameters of the line!
• Next question
• How to find the intersection ?

11
Voting Scheme

• Each edge pixel in the image votes in Hough
  space for
• a series of
• Choose the of maximum votes

12
Basic Hough Transform
13
Example

14
Example

15
Extension

• Choose the sampling of
• Use gradient of the image voting for specific
• Iteratively find the maximum votes and remove
  corresponding edge pixels
• Suppress edge pixels close to the detected lines

16
Example of Using Estimated Edge
OrientationIterative line removal

17
(No Transcript)
18
A detour through scale space
19
Image encoding-decoding

• 1) Image statistics pixel in neighborhood are
  correlated, encode per pixel value is redundant
• 2) Predictive Codinguse raster scan, predict
  based on pass value, and store only the error in
  prediction. Simple and fast

20
Non-causal prediction

• non-causal involves typically transform, or
  solution to a large sets of equations. Encode
  block by block. Bigger compression but slower.

21
Gaussian Pyramid for encoding
Burt Adelson, 1983

1. Prediction using weighted local Gaussian average
2. Encode the difference as the Laplacian
3. Both Laplacian and the Averaged image is easy to
   encode

22
Gaussian pyramid
23
Choice in weighting function
Gaussian
24
Image Expansion
25
Image Expansion
26
(No Transcript)
27
-

Laplaican Image
28
Gaussian pyramid is smoothgt can be subsampled
Laplacian pyramid has narrow band of frequencygt
compressed
29
Ln Gn
30
(No Transcript)
31
(No Transcript)