Scribd
Upload
138 views15 pages

EE 604 Image Processing

This document discusses various image processing techniques including: 1) Spatial averaging which involves convolving an input image with a template image to perform low-pass filtering and reduce noise. 2) Edge detection methods like Sobel, Roberts, and Canny edge detectors which use gradient modeling and thresholding to locate edges in an image. 3) Median filtering which replaces pixel values with the median of neighboring pixels to effectively remove salt and pepper noise from an image.

Uploaded by

muskan agarwal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
138 views15 pages

EE 604 Image Processing

This document discusses various image processing techniques including: 1) Spatial averaging which involves convolving an input image with a template image to perform low-pass filtering and reduce noise. 2) Edge detection methods like Sobel, Roberts, and Canny edge detectors which use gradient modeling and thresholding to locate edges in an image. 3) Median filtering which replaces pixel values with the median of neighboring pixels to effectively remove salt and pepper noise from an image.

Uploaded by

muskan agarwal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 15

EE 604

Image Processing
Spatial averaging

• This involves the convolution of the input image with a ‘template’ image. Equivalently, this
is equivalent to the application of an FIR filter in 2D.
• Output image size is a bit bigger than input, as required by convolution, the peripheral
pixels are ignored.
• Used to LP filter before down sampling, and to reduce noise by averaging.

Template examples
Spatial averaging example

Gonzalez
Directional smoothing

Gonzalez
Sharpening operators

•
Un sharp masking

Gonzalez

Google
images
The gradient image

Different
Laplacian
operators

Roberts

Sobel
Gradient modeling
Locating the edge

• Even for 1D a gradual ‘edge’, one is faced


with the question of finding its precise
location.
• So, one proceeds to find the first derivative
(difference), and ‘locates’ the edge
transition point at the maximum of the
gradient.
• Even better, we look for the zero crossing of
the 2nd difference.
• In 2D, the ‘edge’ is defined as the locus of
the edge points. The profile of the transition
can itself vary along the edge, as well as the
start and end levels.
Sumana Gupta
Edge strength and direction

•
Edge detector examples

• Original image with all vertical, and all


horizontal edges found separately.
• The edge operators of different kinds.

Google Images
Canny Edge detector

Currently, the most popular edge detector


is the Canny edge detector, which is a 2
pass algorithm that applies a combination
of initial
• Gaussian smoothing,
• gradient generation,
• maximum suppression,
• double thresholding and
• hysteresis
to ensure to link up broken edges and not
be affected by noise.
Median filtering

• Median filter is nonlinear function of the input image. Does not smear localized noise like a
linear averaging filter dies (mid image below); instead removes it clean.
• The median filter replaces the center pixel by the median value of the neighbours.
• It is not invertible. Very effective in removing salt and pepper noise.
• Fails completely if the noise is so prevalent that it dominates the image.
Morphological processing

• Binary image objects are treated as subsets of a


square grid, consisting only of pixels = 1.
• The remaining pixels, = 0 are called the
‘background object’

• Structuring element: This is a small pre defined


shape that determines the kind of effects that
will result. Usually, a small square, circle, etc.
Dilation and erosion

You might also like