Spatial filtering for Smoothing
For blurring/noise reduction;
Blurring is usually used in preprocessing steps,
e.g., to remove small details from an image prior to object
extraction, or to bridge small gaps in lines or curves
Equivalent to Low-pass filtering in frequency domain because
smaller (high frequency) details are removed based on
neighborhood averaging (averaging filters)
Implementation: simplest form: averaging with a square mask
(assume m×m mask) with the all coefficients equal
to 1/m2 (averaging).
Applications: Reduce noise; smooth false contours
Side effect: Edge blurring
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 1
Smoothing filters
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 2
Spatial filtering for Smoothing (example)
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 3
Spatial filtering for Smoothing (example)
Original image Smoothed by
size: 500 x 500 3 x 3 box filter
Smoothed by Smoothed by
5 x 5 box filter 9 x 9 box filter
Smoothed by Smoothed by
15 x 15 box filter 35 x 35 box filter
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 4
Order-statistics filtering
Nonlinear spatial filters
Output is based on order of gray levels in mask (subimage)
Examples: Median filtering, Max & Min filtering
Median filtering
Assigns the mid value of all the gray levels in the mask to the
center of mask;
Particularly effective when
– noise pattern consists of strong, spiky components (impulse
noise, salt-and-pepper
– edges are to be preserved
– Force pixels with distinct gray levels to be more like their
neighbors
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 5
Median Filtering
Output ?
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 6
Median Filtering (example)
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 7
Spatial filtering for image sharpening
Background: To highlight fine detail in an image or to
enhance blurred detail
Applications: Electronic printing, medical imaging,
industrial inspection, autonomous target
detection (smart weapons)......
Foundation:
Blurring/smoothing is performed by spatial averaging
(equivalent to integration)
Sharpening is performed by noting only the gray level
changes in the image that is the differentiation
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 8
Spatial filtering for image sharpening
Operation of Image Differentiation
Enhance edges and discontinuities
De-emphasize areas with slowly varying gray-level
values
Mathematical Basis of Filtering for Image Sharpening
Implementation by mask filtering
First-order and second-order derivatives
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 9
Sharpening Filters: Unsharp Masking
Obtain a sharp image by subtracting a lowpass filtered (i.e., smoothed) image
from the original image:
- =
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 10
Sharpening Filters: Unsharp Masking (cont’d)
If A=1, we get a high pass filter
If A>1, part of the original image is added back to the high pass filtered image.
A=1.4 A=1.9
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 11
Unsharp Masking (cont’d)
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 12
Image sharpening (High-Pass Filters)
Spatial filters whose effect on the output image is
equivalent to preserving or emphasizing its high-
frequency components (e.g., fine details, points, lines,
and edges), i.e. to highlight transitions in intensity within
the image.
Linear HPFs can be implemented using 2D masks with
positive and negative weights / coefficients, (Laplacian, a
simple, isotropic (i.e., rotation invariant) second-order
derivative)
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 13
First-order and second-order derivatives
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 14
Example
2D convolution
(center location only)
The filter factors
into a product of 1D
filters:
Perform convolution
along rows: * =
Followed by convolution
along the remaining column: * =
Example for discrete derivatives
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 16
Example for discrete derivatives
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 17
Various situations encountered for derivatives
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 18
Various situations encountered for derivatives
Ramps or steps in the 1D profile normally characterize
the edges in an image
f ″ is nonzero at the onset and end of the ramp:
produces thin (double) edges
f ′ is nonzero along the entire ramp: produces thick edges
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 19
Various situations encountered for derivatives
Thin lines
Isolated point
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 20
Comparison between f" and f´
f´ generally produce thicker edges in an image
f" have a stronger response to fine detail
f´ generally have a stronger response to a gray-level step
f" produces double response at step changes in gray level
f" responses given similar changes in gray-level values
point > step > line
For image enhancement, f" is generally better suited
than f´
Major application of f´ is for edge extraction;
f´ used together with f" results in impressive
enhancement effect
10/28/21 COMSATS IIT, Lahore -- Special Topics in Image Processing -- CSC857 21
