Computer Vision: filtering

Computer Vision: filtering

Image filtering Hybrid Images, Oliva et al., http://cvcl.mit.edu/hybridimage.htm 1 Image filtering Hybrid Images, Oliva et al., http://cvcl.mit.edu/hybridimage.htm 2 Reading Szeliski, Chapter 3.1-3.2 3 What is an image? 4 Images as functions We can think of an image as a function, f, from R2 to R: f( x, y ) gives the intensity at position ( x, y ) Realistically, we expect the image only to be defined over a rectangle, with a finite range: f: [a,b]x[c,d] [0,1] A color image is just three functions pasted together. We can write this as a vector-valued function: r ( x, y ) f ( x, y ) g ( x, y ) b( x, y )

5 Images as functions snoop 3D view 6 What is a digital image? In computer vision we usually operate on digital (discrete) images: Sample the 2D space on a regular grid Quantize each sample (round to nearest integer) If our samples are apart, we can write this as: f[i ,j] = Quantize{ f(i , j ) } The image can now be represented as a matrix of integer values 7 Image processing An image processing operation typically defines a new image g in terms of an existing image f. We can transform either the domain or the range of f. Range transformation: Whats kinds of operations can this perform? g (x,y) = f (x,y) + 20 8 Image processing

Some operations preserve the range but change the domain of f : What kinds of operations can this perform? g (x,y) = f (-x,y) 9 Image processing Still other operations operate on both the domain and the range of f . 10 Noise Image processing is useful for noise reduction... Common types of noise: Salt and pepper noise: contains random occurrences of black and white pixels Impulse noise: contains random occurrences of white pixels Gaussian noise: variations in intensity drawn from a Gaussian normal distribution 11 Ideal noise reduction Given a camera and a still scene, how can you reduce noise? 12 Ideal noise reduction Given a camera and a still scene, how can you reduce noise? 13

Practical noise reduction How can we smooth away noise in a single image? 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 100 130 110 120 110 0 0 0 0 0 110 90 100 90 100 0 0 0 0 0 130 100 90 130 110 0

0 0 0 0 120 100 130 110 120 0 0 0 0 0 90 110 80 120 100 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 14 Mean filtering 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0 0

0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0 0

0 0 0 0 0 0 0 0 0 0 90 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 15 Mean filtering 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 90 90 90 90 90 0 0 0 0

0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 0 0 0 0 0

90 90 90 90 90 0 0 0 0 0 0 0 0 0 0 0 0 0 0 90 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 16 Mean filtering 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

90 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0 0 0 90

90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0 0 0 0 0 0 0 0 0 0 0

0 90 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 10 20 30 30 30 20 10 0 20 40 60 60 60 40 20 0 30 60 90 90 90 60 30 0 30 50 80 80 90 60 30 0 30 50 80 80 90 60 30 0 20 30 50 50 60 40 20 10 20 30 30 30 30 20 10 10 10 10 0 17 0 0 0

0 Mean filtering When does this work? F x I x N x 1 G x F x 1 F x F x 1 3 1 G x I x 1 I x I x 1 3 N x 3 1 E G x I x 1 I x I x 1 3 When does E(G(x)) = I(x)? 3I x I x 1 I x I x 1 I x 1 2 I x I x 1 0 18 Mean filtering When does this work? 19 Effect of mean filters 20 Cross-correlation filtering Lets write this down as an equation. Assume the averaging window is (2k+1)x(2k+1): We can generalize this idea by allowing different weights for different neighboring pixels:

This is called a cross-correlation operation and written: H is called the filter, kernel, or mask. The above allows negative filter indices. When you implement need to use: 21 Mean kernel Whats the kernel for a 3x3 mean filter? 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0

0 0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0

0 0 0 0 0 0 0 0 0 0 0 90 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 22 Gaussian Filtering A Gaussian kernel gives less weight to pixels further from the center of the window 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 90 90 90 90 90 0 0

0 0 0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 90 0 0 0 0 0 90 90 90 90 0 0 0

0 0 90 90 90 90 90 0 0 0 0 0 0 0 0 0 0 0 0 0 0 90 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 1

2 4 2 1 2 1 This kernel is an approximation of a Gaussian function: What happens if you increase ? 23 = 1 pixel = 5 pixels = 10 pixels = 30 pixels 24 Mean vs. Gaussian filtering 25 Filtering an impulse a b

c d e f g h i 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 26 Convolution A convolution operation is a crosscorrelation where the filter is flipped both horizontally and vertically before being

applied to the image: It is written: Suppose H is a Gaussian or mean kernel. How does convolution differ from crosscorrelation? Suppose F is an impulse function (previous slide) What will G look like? 27 Continuous Filters We can also apply filters to continuous images. In the case of cross correlation: In the case of convolution: Note that the image and filter are infinite. 28 Median filters A Median Filter operates over a window by selecting the median intensity in the window. What advantage does a median filter have over a mean filter? Is a median filter a kind of convolution? 29 Comparison: salt and pepper noise 30

Comparison: Gaussian noise 31 Linear filters: example * 0 0 0 0 1 0 0 0 0 Original = Identical image 32 Source: D. Lowe Linear filters: example * 0 0 0 1 0 0 0 0 0 Original = Shift by 1 pixel 33 Source: D. Lowe

Linear filters: example * 1 1 1 1 1 1 1 1 1 Original = Blur (with a mean filter) 34 Source: D. Lowe Linear filters: example * 0 0 0 0 2 0 0 0 0 - 1 1 1 1 1 1 1 1 1 Original = 35 Sharpening

filter (accentuates edges) Source: D. Lowe Sharpening 36 Source: D. Lowe Sharpening What does blurring take away? = detail smoothed (5x5) original Lets add it back: = + original detail 37 sharpened Sharpen filter

unfiltered filtered 38 Sharpen filter blurred image image scaled impulse unit impulse (identity) Gaussian 39 Laplacian of Gaussian Enhance it!

Recently Viewed Presentations

  • Quadratic Functions

    Quadratic Functions

    Graphing the Parent Quadratic Function f(x) = x2. Create a table of values. The vertex of the parent function f(x) = x2 is (0,0) and the axis of symmetry is x = 0.
  • REALISMO - pessoal.educacional.com.br

    REALISMO - pessoal.educacional.com.br

    Contudo há semelhanças e diferenças entre essas correntes. As semelhanças residem na objetividade, na luta contra o Romantismo e no gosto por descrições minuciosas. CARACTERÍSTICAS DO NATURALISMO As características específicas do Naturalismo resultam da sua aproximação com as diversas ciências...
  • Hinduism 1 - Lake-Sumter State College

    Hinduism 1 - Lake-Sumter State College

    The Lotus The holiest of flowers for the Hindu The lotus is representative of the true soul of an individual—also a symbol of creation. I. Generalizations of Indian Religion B. It is a highly inclusive religious system. ... 1028 (arranged...
  • Behaviorism - California State University, Fresno

    Behaviorism - California State University, Fresno

    Behaviorism and the Teaching of Mathematics. ... discovered by Pavlov that involves repeatedly pairing a neutral stimulus with a response-producing stimulus until the neutral stimulus triggers the same response. Ivan Pavlov's Classical Conditioning. Before Conditioning.
  • Body Image - scott.k12.ky.us

    Body Image - scott.k12.ky.us

    Times New Roman Arial Calibri Default Design Body Image Bellringer Discussion Body Image Eating Disorders More Statistics What is body image? Origins of Body Image Cosmetic Surgery Body Image Is: Men's Image Issues Media Influence The Diet/Fitness Craze We are...
  • 2007 Grocery SuperStudy 2007 Grocery SuperStudy Sales Average

    2007 Grocery SuperStudy 2007 Grocery SuperStudy Sales Average

    Arial MS Pゴシック Wingdings 3 Tahoma AEB-New Microsoft Graph Chart Microsoft Office Excel Chart 2007 Grocery SuperStudy 2007 Grocery SuperStudy Average Weekly Sales Per Item Weekly Movement Per SKU Percent SKU's Making Money Profitability Return on Investment Return on Investment...
  • The World of Work Prepared by Aisha Alshehhi

    The World of Work Prepared by Aisha Alshehhi

    is one of the world's leading oil companies exploiting vast oil and gas reserves in the emirate of Abu Dhabi and planning to increase oil and gas reserves. why you are interested. Because all people prefer to put adnoc oil...
  • How We Breathe  Increase in thoracic cavity volume

    How We Breathe Increase in thoracic cavity volume

    How We Breathe Increase in thoracic cavity volume decreases pressure, creating lower pressure in lungs than outside, allowing air to rush into lungs