Scribd
Upload
42 views

Filters

CS443: Digital Imaging and Multimedia Filters Spring 2008 Ahmed Elgammal Dept. Of Computer Science Rutgers University Outlines What are filters point operations are limited (why) They cannot accomplish tasks like sharpening or smoothing.

Uploaded by

Eo Un
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
42 views

Filters

CS443: Digital Imaging and Multimedia Filters Spring 2008 Ahmed Elgammal Dept. Of Computer Science Rutgers University Outlines What are filters point operations are limited (why) They cannot accomplish tasks like sharpening or smoothing.

Uploaded by

Eo Un
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

CS443: Digital Imaging and Multimedia

Filters
Spring 2008 Ahmed Elgammal Dept. of Computer Science Rutgers University

Outlines
What are Filters Linear Filters Convolution operation Properties of Linear Filters Application of filters Nonlinear Filter Normalized Correlation and finding patterns in images Sources:
Burger and Burge Digital Image Processing Chapter 6 Forsyth and Ponce Computer Vision a Modern approach

What is a Filter
Point operations are limited (why) They cannot accomplish tasks like sharpening or smoothing

Smoothing an image by averaging


Replace each pixel by the average of its neighboring pixels Assume a 3x3 neighborhood:

In general a filter applies a function over the values of a small neighborhood of pixels to compute the result The size of the filter = the size of the neighborhood: 3x3, 5x5, 7x7, , 21x21,.. The shape of the filter region is not necessarily square, can be a rectangle, a circle Filters can be linear of nonlinear

Linear Filters: convolution

Averaging filter

Computing the filter operation


The filter matrix H moves over the original image I to compute the convolution operation We need an intermediate image storage! We need 4 for loops! In general a scale is needed to obtain a normalized filter. Integer coefficient is preferred to avoid floating point operations

Another smoothing filter

Integer coefficient

Ex: linear filter in Adobe photoshop

For a filter of size (2K+1) x (2L+1), if the image size is MxN, the filter is computed over the range:

Types of Linear Filters

Smoothing by Averaging vs. Gaussian


Flat kernel: all weights equal 1/N

Smoothing with a Gaussian


Smoothing with an average actually doesnt compare at all well with a defocussed lens
Most obvious difference is that a single point of light viewed in a defocussed lens looks like a fuzzy blob; but the averaging process would give a little square.

A Gaussian gives a good model of a fuzzy blob

An Isotropic Gaussian
The picture shows a smoothing kernel proportional to x2 + y2 exp 2 2

(which is a reasonable model of a circularly symmetric fuzzy blob)

Smoothing with a Gaussian

Gaussian smoothing
Advantages of Gaussian filtering
rotationally symmetric (for large filters) filter weights decrease monotonically from central peak, giving most weight to central pixels Simple and intuitive relationship between size of and the smoothing. The Gaussian is separable:

Advantage of seperability
First convolve the image with a one dimensional horizontal filter Then convolve the result of the first convolution with a one dimensional vertical filter For a kxk Gaussian filter, 2D convolution requires k2 operations per pixel But using the separable filters, we reduce this to 2k operations per pixel.

10

Separability

2 1 2 1 3 4 1 2 1

3 5 4 11 18 18

3 5 6

11 18 18

65

1 2 1

1 =

1 2 1

2 4 2

1 2 1

2 3 4

3 5 4

3 5 6

=2 + 6 + 3 = 11 = 6 + 20 + 10 = 36 = 4 + 8 + 6 = 18 65

Advantages of Gaussians
Convolution of a Gaussian with itself is another Gaussian
so we can first smooth an image with a small Gaussian then, we convolve that smoothed image with another small Gaussian and the result is equivalent to smoother the original image with a larger Gaussian. If we smooth an image with a Gaussian having sd twice, then we get the same result as smoothing the image with a Gaussian having standard deviation (2)1/2

11

Mathematical Properties of Linear Convolution


For any 2D discrete signal, convolution is defined as:

Properties
Commutativity Linearity

(notice) Associativity

12

Properties
Separability

Nonlinear Filters
Linear filters have a disadvantage when used for smoothing or removing noise: all image structures are blurred, the quality of the image is reduced. Examples of nonlinear filters:
Minimum and Maximum filters

13

Median Filter
Much better in removing noise and keeping the structures

14

Weighted median filter

15

Linear Filters: convolution

Convolution as a Dot Product


Applying a filter at some point can be seen as taking a dot-product between the image and some vector Convoluting an image with a filter is equivalent to taking the dot product of the filter with each image window.
Window
weights Window

weights

Original image

Filtered image

16

Largest value when the vector representing the image is parallel to the vector representing the filter Filter responds most strongly at image windows that looks like the filter. Filter responds stronger to brighter regions! (drawback) Insight: filters look like the effects they are intended to find filters find effects they look like
Window
weights

Ex: Derivative of Gaussian used in edge detection looks like edges

Normalized Correlation
Convolution with a filter can be used to find templates in the image. Normalized correlation output is filter output, divided by root sum of squares of values over which filter lies Consider template (filter) M and image window N:
C=

M(i, j)N(i, j)
i =1 j =1

[ M(i, j) 2 N(i, j)2 ]1/ 2


i =1 j =1 i =1 j =1

Window Template

Original image

Filtered image (Normalized Correlation Result)

17

Normalized Correlation
C=

M(i, j)N(i, j)
i =1 j =1

[ M(i, j) 2 N(i, j)2 ]1/ 2


i =1 j =1 i =1 j =1

This correlation measure takes on values in the range [0,1] it is 1 if and only if N = cM for some constant c so N can be uniformly brighter or darker than the template, M, and the correlation will still be high. The first term in the denominator, M2 depends only on the template, and can be ignored The second term in the denominator, N2 can be eliminated if we first normalize the grey levels of N so that their total value is the same as that of M - just scale each pixel in N by M/ N

Positive responses Zero mean image, -1:1 scale Zero mean image, -max:max scale

18

Positive responses Zero mean image, -1:1 scale Zero mean image, -max:max scale

Figure from Computer Vision for Interactive Computer Graphics, W.Freeman et al, IEEE Computer Graphics and Applications, 1998 copyright 1998, IEEE

19

You might also like