Arithmetic Coding

Example
 Arithmetic Vs. Huffman Coding

The most common statistical compression methods are Huffman and

Arithmetic coding.

Huffman utilizes a static table to represent all the characters and their
frequencies, then generates a code table accordingly.

More frequent characters will be assigned shorter code,

and by doing so the source can be effective compressed.

Arithmetic coding works a bit differently from Huffman.

It also uses a statistical table for coding, but this table is
Adaptive:
it is modified from time to time to reflect the real time distribution statistics.

While a new character is being processed, the table will re-calculate

frequencies until the end of the text stream. 7
 Arithmetic Vs. Huffman Coding

Huffman uses a static table for the whole coding process, so it is rather fast,
but does not produce an efficient compression ratio.

Arithmetic coding, on the other hand, has different features. It can

generate a high compression ratio, but all the complex calculation takes
much more time, resulting in a slower implementation.

The table below presents a simple comparison between these compression

methods.
Compression Method Arithmetic Huffman
Compression Ratio Very Good Poor
Compression Speed Slow Fast
Decompression Speed Slow Fast
Memory Space Very Low Low
Compressed Pattern Matching No Yes
Permits Random Access No Yes
8
 Arithmetic Vs. Huffman Coding

An ideal compression method should satisfy all those features given in the
table.
Compression Method Arithmetic Huffman
Compression Ratio Very Good Poor
Compression Speed Slow Fast
Decompression Speed Slow Fast
Memory Space Very Low Low
Compressed Pattern Matching No Yes
Permits Random Access No Yes

The last two items are important considerations in information retrieval, as

both features are key in a system's ability to search documents directly and
randomly.
Without compressed pattern matching, a system would need to decompress
the entire document prior to processing a user's query. Without random
access, a system could not retrieve any part of a document until it completely
decompressed the document from the very beginning. 9
 Arithmetic Code
Coding

In arithmetic coding a message is encoded as a number from the

interval [0, 1).

The number is found by expanding it according to the probability of the

currently processed letter of the message being encoded.

This is done by using a set of interval ranges IR determined by the

probabilities of the information source as follows:

IR ={ [0, p1), [p1, p1+ p2), [p1+ p2, p1+ p2+ p3), … [p1+ … + pn-1, p1+ … + pn) }
j

Putting q  p
j i we can write IR = { [0, q1), [q1, q2), …[qn-1, 1) }
i1

10
 Arithmetic Code
Coding

In arithmetic coding these subintervals also determine the proportional

division of any other interval [L, R) contained in [0, 1) into subintervals
IR[L,R] as follows:
IR[L,R] = { [L, L+(R-L) q1), [L+(R-L) q1, L+(R-L) q2), [L+(R-L) q2, L+(R-L) q3), … , [L+(R-L) Pn-1, L+(R-L) ) }

Using these definitions the arithmetic encoding is determined by the

Following algorithm:

ArithmeticEncoding ( Message )
1. CurrentInterval = [0, 1);
While the end of message is not reached
2. Read letter xi from the message;
3. Divid CurrentInterval into subintervals IRCurrentInterval;
Output any number from the CurrentInterval (usually its left boundary);

This output number uniquely encoding the input message.

11
 Arithmetic Code
Coding

Example 1 Consider the information source

A B C #
0.4 0.3 0.1 0.2

Then the input message ABBC#

has the unique encoding number 0.23608.

As we will see the explanation In the next slides

 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

1. CurrentInterval = [0, 1);

Xi Current interval Subintervals

A [0, 1)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

3. Divid CurrentInterval into subintervals IRCurrentInterval;

Xi Current interval Subintervals

A [0, 1)
IR[0,1)= {
[0, 0.4) , [0.4, 0.7),
[0.7, 0.8), [0.8, 1)
}
j

q  p
j i
i1

[L+(R-L) qi, L+(R-L) qi+1)

 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

3. Divid CurrentInterval into subintervals IRCurrentInterval;

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)

IR ={
[0,1)
[0, 0.4) , [0.4, 0.7),
[0.7, 0.8), [0.8, 1)
}
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

Example 1
input message: A B B C #
No. 1

A B C #
0.4 0.3 0.1 0.2

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
[0, 0.4)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

3. Divid CurrentInterval into subintervals IRCurrentInterval;

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4)
IR[0,0.4)= {
[0, 0.16) , [0.16, 0.28),
[0.28, 0.32), [0.32, 0.4)
}
j

q  p j i
i1

[L+(R-L) qi, L+(R-L) qi+1)

 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

3. Divid CurrentInterval into subintervals IRCurrentInterval;

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)

IR[0,0.4)= {
[0, 0.16) , [0.16, 0.2 8),
[0.28, 0.32), [0.32, 0.4)
}
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

Example 1
input message: A B B C #
No. 2

A B C #
0.4 0.3 0.1 0.2
Xi Current interval Subintervals
A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
[0.16, 0.28)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

3. Divid CurrentInterval into subintervals IRCurrentInterval;

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
B [0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

Example 1
input message: A B B C #
No. 2

A B C #
0.4 0.3 0.1 0.2
Xi Current interval Subintervals
A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
B [0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28)

[0.208, 0.244)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

3. Divid CurrentInterval into subintervals IRCurrentInterval;

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
B [0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28)

C [0.208, 0.244) [0.208, 0.2224) , [0.2224, 0.2332), [0.2332, 0.2368), [0.2368, 0.244)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

Example 1
input message: A B B C #
No. 3

A B C #
0.4 0.3 0.1 0.2
Xi Current interval Subintervals
A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
B [0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28)

C [0.208, 0.244) [0.208, 0.2224) , [0.2224, 0.2332), [0.2332, 0.2368), [0.2368, 0.244)

[0.2332, 0.2368)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

3. Divid CurrentInterval into subintervals IRCurrentInterval;

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
B [0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28)

C [0.208, 0.244) [0.208, 0.2224) , [0.2224, 0.2332), [0.2332, 0.2368), [0.2368, 0.244)

# [0.2332, 0.2368) [0.2332, 0.23464) , [0.23464, 0.23572), [0.23572, 0.23608), [0.23608, 0.2368)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #
No. 4

A B C #
0.4 0.3 0.1 0.2
Xi Current interval Subintervals
A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
B [0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28)

C [0.208, 0.244) [0.208, 0.2224) , [0.2224, 0.2332), [0.2332, 0.2368), [0.2368, 0.244)

# [0.2332, 0.2368) [0.2332, 0.23464) , [0.23464, 0.23572), [0.23572, 0.23608), [0.23608, 0.2368)

[0.23608, 0.2368)
 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

2. Read Xi

Example 1
input message: A B B C #

# is the end of input message Stop Return current interval [0.23608, 0.2368)

Xi Current interval Subintervals

A [0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
B [0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4)
B [0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28)

C [0.208, 0.244) [0.208, 0.2224) , [0.2224, 0.2332), [0.2332, 0.2368), [0.2368, 0.244)

# [0.2332, 0.2368) [0.2332, 0.23464) , [0.23464, 0.23572), [0.23572, 0.23608), [0.23608, 0.2368)

[0.23608, 0.2368)
 A B C #
Example 1 ABBC# 0.4 0.3 0.1 0.2

After B C #
A B
seeing
1 0.4 0. 28 0.244 0.2368

0.8 0.23608
0.7

0.4

0 0 0.16 0.208 0.2332

 Arithmetic Code A B C #

Coding 0.4 0.3 0.1 0.2

Example 1
input message: A B B C #

# is the end of input message Stop Return current interval [0.23608, 0.2368)

Return the lower bound of the currentinterval as the codeword

of the input message

Input message Codeword

ABBC# 0.23608
Example 1 A B C #
• The size of the final range is 0.4 0.3 0.1 0.2
0.2368 - 0.23608 = 0.00072,

that is also exactly the multiplication of the probabilities of

the five symbols in the message ABBC#
:
(0.4)* (0.3)* (0.3)* (0.1)* (0.2) = 0.00072.

it takes 5 decimal digits to encode the message.

• According to Shannon :
The best compression code is the output length contains a contribution
of –log(p) bits from the encoding of each symbol whose probability of
occurrence is p.

The entropy of ABBC# is :

-log 0.4 -log 0.3 -log 0.3 -log 0.1 -log 0.2 =-log 0.00072 = 3.14
 Arithmetic Code
Decoding

Arithmetic decoding can be determined by the following algorithm:

ArithmeticDecoding ( Codeword )
1. CurrentInterval = [0, 1);
While(1)
2. Divid CurrentInterval into subintervals IRCurrentInterval;
3. Determine the subintervali of CurrentInterval to which
Codeword belongs;
4. Output letter xi corresponding to this subinterval;
5. If xi is the symbol ‘#’
Return;
6. CurrentInterval = subintervali in IRCurrentInterval;
 Arithmetic Code
Decoding

Example Consider the information source

Symbol Probability
A 0.4
B 0.3
C 0.1
# 0.2

Then the input code word 0.23608 can be decoded

to the message ABBC#

As we will see the explanation In the next slides

 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

0. CurrentInterval = [0, 1);

Current interval Subintervals Output

[0, 1)
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

1. Divid CurrentInterval into subintervals IRCurrentInterval;

Current interval Subintervals Output

[0, 1) IR[0,1)= {
[0, 0.4) , [0.4, 0.7),
[0.7, 0.8), [0.8, 1)
}
j

q  p
j i
i1

[L+(R-L) qi, L+(R-L) qi+1)

 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)

IR[0,1)= {
[0, 0.4) , [0. 4, 0.7),
[0.7, 0.8), [0. 8, 1)
}
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

2. Determine the subintervali of CurrentInterval to which Codeword belongs;

0 ≤ 0.23608 < 0.4

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

2. Determine the subintervali of CurrentInterval to which Codeword belongs;

0 ≤ 0.23608 < 0.4

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1)
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

3. Output letter xi corresponding to this subinterval;
No. 1
A B C #
No. 1
0.4 0.3 0.1 0.2
Current interval Subintervals Output
[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1) A
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

4. If xi is the symbol ‘#’

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1) A
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

4. If xi is the symbol ‘#’ NO

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1) A
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

5. CurrentInterval = subintervali in IRCurrentInterval;

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1) A
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

5. CurrentInterval = subintervali in IRCurrentInterval;

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1) A
[0, 0.4)
 Arithmetic Code A B C #

Decoding 0.4 0.3 0.1 0.2

Example input codeword: 0.23608

Similarly we repeat the algorithm steps 1 to 5 until the output symbol = ‘#’

Current interval Subintervals Output

[0, 1) [0, 0.4) , [0.4, 0.7), [0.7, 0.8), [0.8, 1) A
[0, 0.4) [0, 0.16) , [0.16, 0.28), [0.28, 0.32), [0.32, 0.4) B
[0.16, 0.28) [ 0.16, 0.208) , [0.208, 0.244), [0.244, 0.256), [0.256, 0.28) B
[0.208, 0.244) [0.208, 0.2224) , [0.2224, 0.2332), [0.2332, 0.2368), [0.2368, 0.244) C
[0.2332, 0.2368) [0.2332, 0.23464) , [0.23464, 0.23572), [0.23572, 0.23608), [0.23608, 0.2368) #

4. If xi is the symbol ‘#’ Yes Stop

Return the output message: A B B C #