Image Compression
Why Do We Need Compression?
• Requirements may outstrip the anticipated
increase of storage space and bandwidth
• For data storage and data transmission
– DVD
– Video conference
• The bit rate of uncompressed digital cinema
data exceeds 1 Gbps
Why Can We Compress?
• Spatial redundancy
– Neighboring pixels are not independent but
correlated
• Temporal redundancy
Information vs Data
REDUNDANT
DATA
INFORMATION
DATA = INFORMATION + REDUNDANT DATA
An Image Model-Ref: J.B.O’Neal
Picture size is one unit wide by one unit high
Width M=Number of
1 Unit Samples
D=Spacing Between
1
Samples =
1/2 M
Height M Correlation Between
1 Unit D
Adjacent Samples
D
e e
M
M1/2
Compression As It Relates To Image Content
Picture Correlation Distance -1
•Portrait 6.3
(Fills 1/2 Frame)
•Typical 16.7
(Moderate Detail)
•100 People 50
•2000 People 150
INTERFRAME and INTRAFRAME
PROCESSING
Interframe Processing Point to Point
Predictive Encoding
Line to Line
Intraframe
Processing
BIT RATE = NQF
N = NUMBER OF PIXELS
Q = QUANTIZATION BITS/PIXEL
F = FRAME RATE
Channel Bit Rate
Compression Ratio = 10 LOG
NQF
REDUCING CREATES
N Reduced Resolution
Q Contouring (Artifacts)
F Image Blur
We need More Sophisticated Approaches
Selected Methods for
compression
• LPC
• Delta Modulation
• Bit plane encoding
• Transform encoding
• Standards
– JPEG
– MPEG
PREDICTIVE CODING
• Predictive Coding
transmit the difference between
estimate of future sample &
the sample itself.
- Delta modulation
- DPCM
- Adaptive predictive coding
- Differential frame coding
Differential Pulse Code Modulation
(DPCM)
SIMPLE DELTA
MODULATION
(t) +
x(t)
+ -
y (t)
f (t)
SIMPLE DELTA MODULATION
y(t)
x(t)
f(t)
t
BIT PLANE ENCODING
abcdef
a
b
c
d
e
f
TRANSFORM CODING
• Transform Coding
- transform image
- code the coefficients of the transform
- transmit them
- reconstruct by inverse transform
• Benefits
- transform coeff. relatively uncorrelated
- energy is highly compacted
- reasonable robust relative to
channel errors
Potential Bit Rate Reduction
for 525 Line Video Imagery
TYPE OPERATIONS COMMENTS
Decrease N Resolution 4:1
(from 512 to 256)
Decrease Q Dynamic Range 6:5
of Display
Decrease F Frame Rate 6:1
Flicker Results (30 Fps 5 Fps)
Image Blurring
Delta Point to Point 2:1 3:1
Modulation Redundancy 6 Bits/Pixel to
2 Bits/Pixel
TYPE OPERATIONS COMMENTS
Adaptive Intraframe 5:1
Intraframe 6 Bits/Pixel to
Redundancy
PCM - Delta 1.2 Bits/Pixel
Optimal 12:1
Transform Intraframe
6 Bits/Pixel to
Encoding Redundancy
1.2 Bits/Pixel
JPEG Intraframe
27:1
Redundancy
Interframe Frame to Frame Hundreds to
Processing Redundancy one
COMPRESSION/COST RATIO RANKING
Compression/ Compression/ Vs.
RANK Technique
Cost Ratio 6-Bit PCM
1 “D” MOD 3.0 3.0
1 “D” Adaptive 1.1 3.3
MOD
2 “D” W-H 0.454 12.7
Transform
1 “D” DPCM 0.429 3.0
Hybrid Techniques
• Delta Modulation of transform coefficients
• Variable scan rate techniques
• Contour encoding
• JPEG and MPEG
Lossless or Lossy Compression
• Lossless compression
– There is no information loss, and the image can be
reconstructed exactly the same as the original
– Applications: Medical imagery, Archiving
• Lossy compression
– Information loss is tolerable
– Many-to-1 mapping in compression eg. quantization
– Applications: commercial distribution (DVD) and rate
constrained environment where lossless methods can
not provide enough compression ratio
Standards
• JPEG
• MPEG
Why Do We Need International
Standards?
• International standardization is conducted to
achieve inter-operability .
– Only syntax and decoder are specified.
– Encoder is not standardized and its
optimization is left to the manufacturer.
• Standards provide state-of-the-art
technology that is developed by a group of
experts in the field.
– Not only solve current problems, but also
anticipate the future application requirements.
What Is JPEG?
• "Joint Photographic Expert Group". Voted
as international standard in 1992.
• Works with color and grayscale images,
e.g., satellite, medical, ...
• Lossy and lossless
JPEG
(Intraframe coding)
• First generation JPEG uses DCT+Run
length Huffman entropy coding.
• Second generation JPEG (JPEG2000) uses
wavelet transform + bit plane coding +
Arithmetic entropy coding.
Why DCT Not DFT?
• DCT is similar to DFT, but can provide a better
approximation with fewer coefficients
• The coefficients of DCT are real valued instead of complex
valued in DFT.
The 64 (8 X 8) DCT Basis
Functions
• Each 8x8 block can
be looked at as a
weighted sum of
these basis functions.
• The process of 2D
DCT is also the
process of finding
those weights.
Zig-zag Scan DCT Blocks
• Why? -- To group low frequency
coefficients in top of vector.
• Maps 8 x 8 to a 1 x 64 vector.
Original
JPEG
27:1
JPEG2000
27:1
JPEG Compression Example
• Original image
– 512 x 512 x 8 bits
= 2,097,152 bits
• JPEG
– 27:1 reduction
=77,673 bits
