You are on page 1of 40

EE412/CS455

Principles of Digital Audio and


Video

Lecture 16

Image Enhancement Techniques


Problem: Removal of noise in images from
Space telescope!

Image Averaging

Averaging a set of same pictures overtime

e.g:
Space Telescope Imagery

Example: Image Averaging


Original,
noisy and
average
images
(using
8,16,64,128
noisy
images)

DIFFERENCE IMAGES AND THEIR


HISTOGRAMS
Original - Average
images (using
8,16,64,128 noisy
images)

Spatial Filtering

Linear Spatial Filtering

Convolving an image by a mask (filter,


kernel, template etc.

Non-Linear Spatial Filtering

Spatial Filtering: 3x3 Mask

Choose a Mask For:

Smoothing
Detection of edges

Smoothing Spatial Filtering

Averaging Filter
A linear filter for blurring and noise removal
e.g: Box filter, Weighted average filter
=> Averaging filter is a low pass filter
Use: gross representation of the image

Median Filter
A nonlinear filter for salt-and-pepper noise removal
variants: max and min filter

Example: Smoothing Spatial


Filtering
Original and
result of
smoothing
with square
averaging
filter of sizes
3, 5, 9, 15,
25, 45, 55

Example: Averaging and


Thresholding

Example: Median Filtering

Original

Averaging mask filtered

Median

Edge Detection and Sharpening

Derivative Example

Edge Detection and Sharpening


The Gradiant operators : Based on first order derivatives
and used primarily for edge detection

Roberts cross-gradiant operators


Two masks of sizes 2x2

Prewitt operators
Two masks of sizes 3x3

Sobel operators
Two masks of sizes 3x3

Gradiant Masks

Example: Sobel Filter

Sharpening Spatial Filtering

More types are also available like the second


order Laplacian operators
(Study Gradient operator)

Tips on Mixing and Effect


Processing

Mixing and Effect Processing in


Audio

Audio Mixing
Combining multiple audio signals into one
or more (composite) signal

Capabilities of level control, equalization,


introduction of effects

Professional Audio Mixing Example

Professional Audio Equalization


Example

From: http://upload.wikimedia.org/wikipedia/commons/8/82/Graphic_equalizer.jpg

Mixing and Effect Processing in


Images, Video and Graphics

Video Mixing
Combining multiple video signal into a single
output video stream

Video Effects Processing

Most 2D effects can be implemented


by processing the two dimensions
separately

Skewing Example

Windowing and Zooming

Chroma Keying

Replace one part of the image with another

Chroma Keying

HW and Computational
Complexity

Computing of Images/Video

Processing of Images/Videos is a very


computation intensive task

=> There is a need of special hardware for it

Digital Signal Processors (DSP)

Special class of processors designed for high


speed signal processing by supporting rapid
movement of Data and providing special DSP
instructions
Examples:
Trimedias TM1300, TM1500..
Texas Instruments TMS320DM64x (720 MHz)
etc..

Digital Signal Processors (DSP)


Key Features
General Purpose (Digital Signal
Processing/Media Processing)
Special Instructions for DSP operations:
MAC, interpolation, decimation etc.
Separate Memories (for Program and Data)
Separate Buses
RISC
Heavily pipelined
etc.

Pipelining

ASICs

Application Specific Integrated Circuits


(ASICs) for dedicated tasks
Examples:
MPEG encoders/ decoders

Memory

Memory is always organized linearly even


in case of video

Memory

Even video memory has a linear structure

Frame Memory Word Structure

Necessary Readings
Reading Package
Page 251-254
Page 262-274
Page 260-261 (section 8.3.3,8.3.3.1,8.3.3.2)
Page 281-314 (Gonzalez)

You might also like