computer vision, winter 2012-13 cs department, technion

17
Image Pyramids Computer Vision, winter 2012-13 CS Department, Technion

Upload: alexis-bell

Post on 11-Jan-2016

218 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Computer Vision, winter 2012-13 CS Department, Technion

Image Pyramids

Computer Vision, winter 2012-13CS Department, Technion

Page 2: Computer Vision, winter 2012-13 CS Department, Technion

TopicsThe Gaussian PyramidThe Laplacian PyramidApplications:

Pattern MatchingCoding (Compression)EnhancementBlending

Page 3: Computer Vision, winter 2012-13 CS Department, Technion

Gaussian Pyramid

Page 4: Computer Vision, winter 2012-13 CS Department, Technion

Gaussian PyramidApproximation filter is not Gaussian

A 5 tap binomial kernel:

2D extension:

1[1,4,6,4,1]

16

1 4 6 4 1

4 16 24 16 41

6 24 36 24 6256

4 16 24 16 4

1 4 6 4 1

Page 5: Computer Vision, winter 2012-13 CS Department, Technion

Pattern Matching

Page 6: Computer Vision, winter 2012-13 CS Department, Technion

Laplacian Pyramid

Page 7: Computer Vision, winter 2012-13 CS Department, Technion

Laplacian PyramidApproximation of the Laplacian of each scaleConstructed by Difference of Gaussian (DoG)

layers

Page 8: Computer Vision, winter 2012-13 CS Department, Technion

Laplacian Pyramid

Page 9: Computer Vision, winter 2012-13 CS Department, Technion

Image Coding

Page 10: Computer Vision, winter 2012-13 CS Department, Technion
Page 11: Computer Vision, winter 2012-13 CS Department, Technion

Enhancement

Page 12: Computer Vision, winter 2012-13 CS Department, Technion

Enhancement

Page 13: Computer Vision, winter 2012-13 CS Department, Technion

Pyramid Blending

Page 14: Computer Vision, winter 2012-13 CS Department, Technion

Step I - Create Blend Masks

Page 15: Computer Vision, winter 2012-13 CS Department, Technion

Step II

• Weighted sum over each Laplacian pyramid scale (of the images) independently

• Gaussian pyramid of masks, or simple resizing of masks to correct dimensions

Blend 1Blend 1 MaskBlend 2Blend 2 Mask

Page 16: Computer Vision, winter 2012-13 CS Department, Technion

Step IIIStandard Laplacian pyramid

reconstruction of blended scales

Page 17: Computer Vision, winter 2012-13 CS Department, Technion

Implementation IssuesImage registrationFeatheringOver/under flows outside [0,1]MATLAB function: impyramid()

Only Gaussian pyramid‘reduce’ or ‘expand’ modes

MATLAB function: imresize()Useful for masks resizing