Image Restroation
Uploaded by
dilip saiImage Restroation
Uploaded by
dilip saiDigital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
The size of typical still image (1200x1600)
1200 1600 3byte 5760000byte
5,760Kbyte 5.76Mbyte
The size of two hours standard television (720x480)
movies
frame pixels bytes
30 (760 480) 3 31,104,000bytes / sec
sec frame pixel
bytes sec
31,104,000 (60 60) 2hours 2.24 1011bytes
sec hour
224GByte.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Data, Information, and Redundancy
• Information
• Data is used to represent information
• Redundancy in data representation of an
information provides no relevant information or
repeats a stated information
• Let n1, and n2 are data represents the same
information. Then, the relative data redundancy R
of the n1 is defined as
R = 1 – 1/C where C = n1/n2
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
• Redundancy in Digital Images
– Coding redundancy
usually appear as results of the uniform
representation of each pixel
– Spatial/Temopral redundancy
because the adjacent pixels tend to have
similarity in practical.
– Irrelevant Information
Image contain information which are ignored
by the human visual system.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Coding Redundancy Spatial Redundancy Irrelevant Information
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Coding Redundancy
• Assume the discrete random variable for rk in the interval
[0,1] that represent the gray levels. Each rk occurs with
probability pk
• If the number of bits used to represent each value of rk by
l(rk) then L1
Lavg k l(rk )
0
• The average code p(rbits
k ) assigned to the gray level values.
• The length of the code should be inverse proportional
to its probability (occurrence).
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Examples of variable length encoding
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Spatial/Temopral Redundancy
• Internal Correlation between the pixel result from
Respective Autocorrelation
– Structural Relationship
– Geometric Relation ship
• The value of a pixel can be reasonably predicted
from the values of its neighbors.
• To reduce the inter-pixel redundancies in an image
the 2D array is transformed (mapped) into more
efficient format (Frequency Domain etc.)
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Irrelevant information and Psycho-Visual
Redundancy
• The brightness of a region depend on other factors
that the light reflection
• The perceived intensity of the eye is limited an non
linear
• Certain information has less relative importance that
other information in normal visual processing
• In general, observer searches for distinguishing
features such as edges and textural regions.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Measuring Information
• A random even E that occurs with
probability P(E) is said to contain I(E)
information where I(E) is defined as
I(E) = log(1/P(E)) = -log(P(E))
• P(E) = 1 contain no information
• P(E) = ½ requires one bit of
information.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Measuring Information
• For a source of events a0, a1,a2, .., ak with
associated probability P(a0), P(a1), P(a2), .., P(ak).
• The average information per source (entropy) is
k
H j 0 P(a ) log(P(a )
j j
For image, we use the normalized histogram to generate
the source probability, which leas to the entropy
L1
~
H pr (ri ) log(pr
i
0
(ri ))
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Fidelity Criteria
• Objective Fidelity Criteria
– The information loss can be expressed as a function
of encoded
the and decoded images.
– For image I (x,y) and its decoded approximation I’(x,y)
– For any value of x and y, the error e(x,y) could be defined as
e(x, y) I '(x, y) I (x, y)
– For the entire Image
M 1 N 1
x0 y I '(x, y) I (x, y)
0
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Fidelity Criteria
• The mean-square-error, erms is
M 1 N 1
I '(x, y) I (x,
erms
y)
The mean-square-error signal-to-noise ratio SNR
2
ms is
x0 y 0
M 1 N 1
SNRms
x0 y
0
I '(x, y) 2
M 1 N 1
I '(x, y) I (x,
y)2
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Three approximations of the same image
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Huffman coding is an entropy
encoding algorithm used for
lossless data compression. The
term refers to the use of a variable-
length code table for encoding a
source symbol (such as a character
in a file) where the variable-length
code table has been derived in a
particular way based on the
estimated probability of occurrence
for each possible value of the
source symbol.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Huffman coding
Assignment procedure
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Arithmetic coding is a form of
variable-length entropy encoding.
A string is converted to
arithmetic encoding, usually
characters are stored with fewer
bits
Arithmetic coding encodes the
entire message into a single
number, a fraction n where (0.0
≤ n < 1.0).
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Compression
Algorithms
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Symbol compression
This approaches determine a set of
symbols that constitute the image,
and take advantage of their multiple
appearance. It convert each symbol
into token, generate a token table
and represent the compressed image
as a list of tokens.
This approach is good for document
images.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
ChChapa
ptterer
ImageImage
88
CompressionCompression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
ChChapa
ptterer
ImageImage
88
CompressionCompression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
DFT and DCT
The periodicity implicit in the 1-D DFT
and DCT. The DCT provide better
continuity that the general DFT.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Bock Size vs. Reconstruction Error
The DCT provide the least error at almost any sub-image size.
The error takes its minimum at sub-images of sizes between 16 and 32.
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Lossless Predictive coding
The encoder expects a discrete sampl es
of a signal f(n).
1. A predictor is applied and its
output is rounded to the nearest
integer. fˆ(n)
The decoder uses the predictor and the
2. The error is estimated as
e(n) f (n) fˆ(n) error stream to reconstructs the
3. The compressed stream consist of original signal f(n).
first sample and the errors, 1. The predictor is initialized using
encoded using variable length the first sample.
coding 2. The received error is added to
predictor result.
f (n) fˆ(n)
e(n)
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Lossless Predictive coding
Linear predictors usually have the
form:
m
fˆ (n) ai f (n i)
round i 0
Original Image (view of the earth).
The prediction error and its histogram.
1. The error is small in uniform
regions
2. Large close to edges and sharp
changes in pixel intensities
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Lossy Predictive coding
The encoder expects a discrete
samples of a signal f(n).
1. A predictor is applied and its
output is rounded to the nearest
integer, fˆ(n)
2. The error is mapped into limited
rage of values (quantized) e(n)
3. The compressed stream consist
of first sample and the
mapped errors, encoded using
variable length coding
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd ed.
Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
Prediction Error
The following images show the
prediction error of the
predictor
fˆ(x, y) 0.97 f (x, y 1)
fˆ(x, y) 0.5 f (x, y 1) 0.5 f (x 1,
0.97 f (x, y 1) h v
y)
fˆ(x, y)
ˆ 0.97 f (x 1, y)
f (x,|y)f
h (x 0.75
1, y)f (x,
f (xy 1)
1, 0.75
y 1) f (x
otherwise
1, |
v y) f (x, (x 1,
0.5 yf 1) f y(x1)
1, y 1)
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
ChChapa
ptterer
ImageImage
88
CompressionCompression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
Digital Image Processing, 3rd
ed. Gonzalez & Woods
www.ImageProcessingPlace.com
Chapter 8
Image Compression
© 1992–2008 R. C. Gonzalez & R. E. Woods
You might also like
- Standard Operating Procedure (SOP) For Shipping An Outbound PackageNo ratings yetStandard Operating Procedure (SOP) For Shipping An Outbound Package15 pages
- Updated - MT390 - Tutorial 6 - Spring 2020 - 21No ratings yetUpdated - MT390 - Tutorial 6 - Spring 2020 - 2165 pages
- Neeraj Tripathi: Simple Dynamic D-LatchNo ratings yetNeeraj Tripathi: Simple Dynamic D-Latch14 pages
- ImageCompression-UNIT-V-students MaterialNo ratings yetImageCompression-UNIT-V-students Material88 pages
- Cisco LISP Configuration Guide: Version 3 2 July 2010No ratings yetCisco LISP Configuration Guide: Version 3 2 July 201026 pages
- Image Compression: Mohamed N. Ahmed, PH.DNo ratings yetImage Compression: Mohamed N. Ahmed, PH.D67 pages
- Ikeja Electric PLC's Financial Statement For Statutory ReportNo ratings yetIkeja Electric PLC's Financial Statement For Statutory Report76 pages
- Unit 5: Image Compression: © 2002 R. C. Gonzalez & R. E. WoodsNo ratings yetUnit 5: Image Compression: © 2002 R. C. Gonzalez & R. E. Woods38 pages
- Image Compression: - Data vs. Information - Entropy - Data RedundancyNo ratings yetImage Compression: - Data vs. Information - Entropy - Data Redundancy30 pages
- Hai An Agency & Logistics Co.,LTD (HAAL)No ratings yetHai An Agency & Logistics Co.,LTD (HAAL)26 pages
- K. Palepu - Business Analysis Valuation - Ch.1No ratings yetK. Palepu - Business Analysis Valuation - Ch.140 pages
- Image Compression (Chapter 8) : CS474/674 - Prof. BebisNo ratings yetImage Compression (Chapter 8) : CS474/674 - Prof. Bebis128 pages
- Image Processing and Compression Techniques: Digitization Includes Sampling of Image and Quantization of Sampled ValuesNo ratings yetImage Processing and Compression Techniques: Digitization Includes Sampling of Image and Quantization of Sampled Values14 pages
- Banyuhay: Katutubong Sayaw Sa Makabagong Pananaw PlaybillNo ratings yetBanyuhay: Katutubong Sayaw Sa Makabagong Pananaw Playbill18 pages
- Get EPIC - Design Your Personal Success Plan - 01 (1) GG GuideNo ratings yetGet EPIC - Design Your Personal Success Plan - 01 (1) GG Guide1 page
- (Ebook PDF) Linear Algebra With Applications 9th Edition Download100% (2)(Ebook PDF) Linear Algebra With Applications 9th Edition Download50 pages
