Title: BAE 790I BMME 231 Fundamentals of Image Processing Class 11
1BAE 790I / BMME 231Fundamentals of Image
ProcessingClass 11
- Fourier Domain Filtering
- Low-pass filters Ideal, Hamming, Hanning,
Butterworth, Gaussian - Sharpening Filters Derivative operators,
Laplacians, Unsharp masking
2Fourier-domain Filters
- Filters can be designed in the Fourier domain as
well as the spatial domain
FFT
FFT-1
H(wx,wy)
Objective Determine a suitable form of this
function
3Ideal Low-pass Filter
- Consider the ideal low-pass filter in 1D
4Ideal Low-pass Filter
5Ideal low-pass filter
SNR 21.4 dB
6Commonly-used Fourier Filters
- All are given as isotropic filters
7Commonly-used Fourier Filters
8Commonly-used Fourier Filters
The order N determines rate of cutoff
9Commonly-used Fourier Filters
10Filter Comparison
11Butterworth filterorder 8, cutoff .2 cycles/pix
SNR 21.8 dB
12Sharpening Filters
- Consider taking the directional first derivative
of an image - This is the same as applying a filter with a
transfer function of jwx
13Gradient Operators
- Simple derivatives can be applied as convolution
masks - These take a local slope of intensity.
- Results are the x- and y-components of the
intensity gradient.
0 0 0 -1 1 0 0 0 0
0 -1 0 0 1 0 0 0 0
x-gradient
y-gradient
14Gradient Operators
- Derivatives in other directions can be determined
from x- and y- derivatives. - Are these LSI operations?
- Results of gradient operators are sensitive to
regions of high Dintensity/Dspace, which we
perceive as what?
15Gradient Operators
0 0 0 -1 1 0 0 0 0
0 -1 0 0 1 0 0 0 0
X gradient
Y gradient
16Second Derivative Operators
- Consider second derivatives
- In terms of convolution masks,
0 0 0 -1 2 -1 0 0 0
0 -1 0 0 2 0 0 -1 0
1 -1 0 -1 1 0 0 0 0
17Second Derivative Operators
- The Laplacian
- Rotationally invariant
- Sensitive to regions of high curvature
18Laplacian Operator
0 0 0 0 0 0 0 -1 0 0 0
-1 4 -1 0 0 0 -1 0 0 0 0
0 0 0
Mask
19Unsharp Masking
- Old photographic sharpening technique
- To sharpen an image, take a blurred version and
subtract it from the original. - image low-pass(image)
- Equivalent to taking a high-pass filtered version
and adding it. - image high-pass(image)
20Unsharp Masking
0 0 0 0 0 0 0 -1 0 0 0
-1 5 -1 0 0 0 -1 0 0 0 0
0 0 0
Mask
21Profiles from Unsharp Masking
22Fourier-domain Root Filtering
a .6
a 1.4