Introduction to conventional compression solutions

Storage constraints and bandwidth limitations in communication systems have

necessitated the search for efficient image compression techniques. Uncompressed
multimedia (graphics, audio and video) data requires considerable storage capacity
and transmission bandwidth. Despite rapid progress in mass-storage density, processor
speeds, and digital communication system performance, demand for data storage
capacity and data-transmission bandwidth continues to outstrip the capabilities of
available technologies. The recent growth of data intensive multimedia-based web
applications have not only sustained the need for more efficient ways to encode signals
and images but have made compression of such signals central to storage and
communication technology.

Principles behind Compression

A common characteristic of most images is that the neighboring pixels are correlated
and therefore contain redundant information. The foremost task then is to find less
correlated representation of the image. Two fundamental components of compression
are redundancy and irrelevancy reduction. Redundancy reduction aims at removing
duplication from the signal source (image/video). Irrelevancy reduction omits parts of
the signal that will not be noticed by the signal receiver, namely the Human Visual
System (HVS). In general, three types of redundancy can be identified:

• Spatial Redundancy or correlation between neighboring pixel values.

• Spectral Redundancy or correlation between different color planes or spectral
bands.
• Temporal Redundancy or correlation between adjacent frames in a sequence of
images (in video applications).

Image compression research aims at reducing the number of bits needed to represent
an image by removing the spatial and spectral redundancies as much as possible.

Classes of Compression Techniques

Lossless vs. Lossy Compression

There are basically two types of compression methods: lossy and lossless. Lossy
compression creates smaller files by discarding some information about the original
image. It removes details and color changes it deems too small for the human eye to
differentiate. Lossless compression, on the other hand, never discards any information
about the original file. In lossless compression schemes, the reconstructed image, after
compression, is numerically identical to the original image. An image reconstructed
following lossy compression contains degradation relative to the original. Often this is
because the compression scheme completely discards redundant information. Under
normal viewing conditions, no visible loss is perceived (visually lossless).
Image Compression Process

Typical Image Coder

A typical lossy image compression system consists of three closely connected

components namely (a) Source Encoder (b) Quantizer, and (c) Entropy Encoder.
Compression is accomplished by applying a linear transform to decorrelate the image
data, quantizing the resulting transform coefficients, and entropy coding the quantized
values.

Source Encoder (or Linear Transformer)

Over the years, a variety of linear transforms have been developed which include
Discrete Fourier Transform (DFT), Discrete Cosine Transform (DCT), Discrete Wavelet
Transform (DWT) and many more, each with its own advantages and disadvantages.

Quantizer

A quantizer simply reduces the number of bits needed to store the transformed
coefficients by reducing the precision of those values. Since this is a many-to-one
mapping, it is a lossy process and is the main source of compression in an encoder.
Quantization can be performed on each individual coefficient, which is known as Scalar
Quantization (SQ). Quantization can also be performed on a group of coefficients
together, and this is known as Vector Quantization (VQ). Both uniform and non-uniform
quantizers can be used depending on the problem at hand.

Entropy Encoder

An entropy encoder further compresses the quantized values losslessly to give better
overall compression. It uses a model to accurately determine the probabilities for each
quantized value and produces an appropriate code based on these probabilities so that
the resultant output code stream will be smaller than the input stream. The most
commonly used entropy encoders are the Huffman encoder and the arithmetic encoder,
although for applications requiring fast execution, simple run-length encoding (RLE) has
proven very effective.

It is important to note that a properly designed quantizer and entropy encoder are
absolutely necessary along with optimum signal transformation to get the best possible
compression.
Further Elaboration on Source Encoder

Discrete Cosine Transformation (DCT)

The discrete cosine transform (DCT) helps separate the image into parts (or spectral sub-
bands) of differing importance (with respect to the image's visual quality). The DCT is
similar to the discrete Fourier transform: it transforms a signal or image from the spatial
domain to the frequency domain. With an input image, A, the coefficients for the output
"image," B, are:

The input image is N2 pixels wide by N1 pixels high; A(i,j) is the intensity of the pixel in row i
and column j; B(k1,k2) is the DCT coefficient in row k1 and column k2 of the DCT matrix.
All DCT multiplications are real. This lowers the number of required multiplications, as
compared to the discrete Fourier transform. The DCT input is an 8 by 8 array of integers.
This array contains each pixel's gray scale level; 8 bit pixels have levels from 0 to 255. The
output array of DCT coefficients contains integers; these can range from -1024 to 1023.
For most images, much of the signal energy lies at low frequencies; these appear in the
upper left corner of the DCT. The lower right values represent higher frequencies, and are
often small - small enough to be neglected with little visible distortion.

Wavelets and Image Compression

Wavelets are functions defined over a finite interval and having an average value of
zero. The basic idea of the wavelet transform is to represent any arbitrary function f(t) as
a superposition of a set of such wavelets or basis functions. These basis functions or baby
wavelets are obtained from a single prototype wavelet called the mother wavelet, by
dilations or contractions (scaling) and translations (shifts). The Discrete Wavelet Transform
of a finite length signal x(n) having N components, for example, is expressed by an N x N
matrix.

Wavelet-based Compression

Despite all the advantages of JPEG compression schemes based on DCT namely
simplicity, satisfactory performance, and availability of special purpose hardware for
implementation; these are not without their shortcomings. Since the input image needs
to be “blocked,” correlation across the block boundaries is not eliminated. This results in
noticeable and annoying “blocking artifacts” particularly at low bit rates. Lapped
Orthogonal Transforms (LOT) attempt to solve this problem by using smoothly overlapping
blocks. Although blocking effect are reduced in LOT compressed images, increased
computational complexity of such algorithms do not justify wide replacement of DCT by
LOT.

Over the past several years, the wavelet transform has gained widespread acceptance
in signal processing in general and in image compression research in particular. In many
applications wavelet-based schemes (also referred as subband coding) outperform
other coding schemes like the one based on DCT. Since there is no need to block the
input image and its basis functions have variable length, wavelet coding schemes at
higher compression avoid blocking artifacts. Wavelet-based coding is more robust under
transmission and decoding errors, and also facilitates progressive transmission of images.
Because of their inherent multiresolution nature, wavelet coding schemes are especially
suitable for applications where scalability and tolerable degradation are important.

JPEG (DCT-Based)

JPEG is the image compression standard developed by the Joint Photographic Experts
Group. It works best on natural images (scenes). The discovery of DCT in 1974 was an
important achievement for the research community working on image compression. The
DCT can be regarded as a discrete-time version of the Fourier-Cosine series. It is a close
relative of DFT, a technique for converting a signal into elementary frequency
components. Thus DCT can be computed with a Fast Fourier Transform (FFT) like algorithm
in O(n log n) operations. Unlike DFT, DCT is real-valued and provides a better
approximation of a signal with fewer coefficients. The DCT of a discrete signal x(n),
n=0,1, .. , N-1 is defined as:

In 1992, JPEG established the first international standard for still image compression where
the encoders and decoders are DCT-based. The JPEG standard specifies three modes
namely sequential, progressive and hierarchical for lossy encoding, and one mode of
lossless encoding.

Entropy Coding (EC) achieves additional compression losslessly by encoding the

quantized DCT coefficients more compactly based on their statistical characteristics. The
JPEG proposal specifies both Huffman coding and arithmetic coding. The baseline
sequential codec uses Huffman coding, but codecs with both methods are specified for
all modes of operation. Arithmetic coding, though more complex, normally achieves 5-
10% better compression than Huffman coding.

JPEG2000 (DWT Based)

JPEG 2000, as noted previously, is the next ISO/ITU-T standard for still image coding. JPEG
2000 is based on the discrete wavelet transform (DWT), scalar quantization, context
modeling, arithmetic coding and post compression rate allocation. The DWT is dyadic
and can be performed with either a reversible filter, which provides for lossless coding, or
a non-reversible one, which provides for higher compression but does not do lossless. The
quantizer follows an embedded dead-zone scalar approach and is independent for
each sub-band. Each sub-band is divided into blocks, typically 64x64, and entropy
coded using context modeling and bit-plane arithmetic coding. The coded data is
organized in so called layers, which are quality levels, using the post-compression rate
allocation and output to the code stream in packets. The generated code-stream is
parseable and can be resolution, layer (i.e. SNR), position or component progressive, or
any combination thereof. JPEG 2000 also supports error-resilience, arbitrarily shaped
region of interest, random access, multicomponent images, palletized color, compressed
domain lossless flipping and simple rotation, to mention a few times of the different
algorithms on a Linux workstation. This only gives an appreciation of the involved
complexity.
Features

Smaller File Sizes

JPEG2000 uses wavelet compression technology to outperform the existing JPEG

standard. Wavelets enable lossy or pure lossless compression, while JPEG always imposes
some sort of loss. (Pure lossless compression means that there is no loss of data the
compressed image is pixel-identical to the original image. Lossy compression discards
data that is considered visually less important). For lossy compression at a given quality,
wavelet-compressed files are generally three to five times smaller than a JPEG
compressed file. For a given file size, a wavelet file offers better quality than a JPEG. For
lossless compression, wavelets can cut the file size by at least half its original size, offering
a smaller file than LZW or Zip.

Better Image Quality

Traditional JPEG compression uses the Discrete Cosine Transformation (DCT), which
compresses an image in 8x8 blocks and results in visible artifacts at high compression
rates. JPEG artifacts include visible seams at the tile edges, dubbed as "blocking
artifacts". The wavelet transform encodes an image in a continuous stream allowing it to
avoid the artifacts that result from DCT’s division of an image into discrete compression
blocks. Wavelet artifacts take the form of blurring high contrast lines, merely making the
image look softer. The wavelet transform performs what's called, multi-resolution
compression—it stores image information in a series of bands, with the most important
image information at the beginning of the file. Each band contains a representation of
the entire image, with the various bands containing details of the image at every level,
from coarse resolution and textures to fine details.

Progressive Image Downloading

An inherent benefit of the wavelet's multi-resolution architecture, is the ability to

progressively access the encoded image in a smooth continuous fashion without having
to download, decode and/or print the entire file. Wavelet compressed images appear
first as an image with coarse resolution and then finer resolution details are progressively
filled in. Since the most important details are stored at the front of the image file, users will
first see a blurry version of the image and the remaining details appearing as the
bitstream arrives. Usually with about 10% of the image data, the user can tell what the
image will be and can decide whether or not to wait for higher resolution. The current
JPEG is single-resolution, so with 10% of the data, the user will have barely gotten a peek
at the top of the image and has wait for the entire download.

Entropy Encoding

Lossless compression techniques frequently involve some form of entropy encoding and
are based in information theoretic techniques. Entropy encoding is an example of lossless
encoding as the decompression process regenerates the data completely. The raw data
and the decompressed data are identical, no information is lost. Entropy encoding just
manipulates bit streams without regarding what the bits mean. It usually transforms the bit
pattern into a different form for transmitting. These methods make use exclusively of the
redundancy of the data. There are several of these kinds. Entropy encoding is used
regardless of the media’s specific characteristics. The data stream to be compressed is
considered to be a simple digital sequence, and the semantic of the data is ignored.
Run Length Encoding (RLE)

Run length coding is an example of entropy encoding. If a byte occurs at least four
consecutive times the number of occurrences is counted. The compressed data
contains this byte followed by a special flag, called M-byte, and the number of its
occurrences. The exclamation mark “!” can be defined as this M-byte. A single
occurrence of this exclamation mark is interpreted as M-byte during the decompression;
two consecutive exclamation marks are interpreted as an exclamation mark occurring
within the data.

This algorithm is very easy to implement and does not require much CPU horsepower. RLE
compression is only efficient with files that contain lots of repetitive data. These can be
text files if they contain lots of spaces for indenting but line-art images that contain large
white or black areas are far more suitable. Computer generated colour images (e.g.
architectural drawings) can also give fair compression ratios.

RLE compression can be used in the following file formats:

• TIFF files
• PDF files
• BMP files
• PCX files

Huffman Encoding

Huffman encoding is an example of entropy encoding. It is based on statistical methods.

Given the character that must be encoded, together with the probability of their
occurrences, the Huffman encoding algorithm determines the optimal code using the
minimum number of bits. Hence the length (number of bits) of the coded characters will
differ. In text, the shortest code is assigned to those characters that occur most
frequently. To determine a Huffman code, it is useful to construct a binary tree. The nodes
of this tree represent the characters that are to be encoded. Every node contains the
occurrence probability of one of the characters belonging to this subtree. 0 and 1 are
assigned to the edges of the tree. The two characters with the lowest probabilities are
combined in the first binary tree. Their root node is labeled with these characters and the
combined probability. The edges are labeled with 1 and 0 respectively. (This assignment
is arbitrary; therefore, with the same data one can get different Huffman codes). The
nodes below this root node won’t be considered anymore. Again, the two nodes with
the lowest probabilities are combined into a binary subtree, the root node and the
edges are labeled as before. Continue in this way until the node with the whole used
alphabet and the probability 1 is reached. The encoded data for the characters are the
paths through the tree to their nodes. The result is stored in a table. If the information of
an image can be transformed into a bit stream, such a table can be used to compress
the data without loss.

This compression algorithm is mainly efficient in compressing text or program files. Images
like they are often used in prepress are better handled by other compression algorithms.

Huffman compression is mainly used in compression programs like pkZIP, lha, gz, zoo and
arj. It is also used within JPEG and MPEG compression.
Arithmetic Coding

In arithmetic coding, a message is encoded as a real number in an interval from one to

zero. Arithmetic coding typically has a better compression ratio than Huffman coding, as
it produces a single symbol rather than several separate code words. Arithmetic coding
is a lossless coding technique. There are a few disadvantages of arithmetic coding. One
is that the whole codeword must be received to start decoding the symbols, and if there
is a corrupt bit in the codeword, the entire message could become corrupt. Another is
that there is a limit to the precision of the number which can be encoded, thus limiting
the number of symbols to encode within a codeword. There also exist many patents
upon arithmetic coding, so the use of some of the algorithms also call upon royalty fees.

Arithmetic coding is used within JPEG2000 compression.

Lempel-Ziv-Welch Algorithm (LZW)

Lempel-Ziv-Welch is a sophisticated lossless compression method which analyzes the

data and looks for repeating patterns. If LZW sees "010101" (for example), it is a clever-
enough algorithm to spot the trend of alternating characters and replace each instance
with a single character, thereby compressing the information. LZW compression replaces
strings of characters with single codes. It does not do any analysis of the incoming text.
Instead, it just adds every new string of characters it sees to a table of strings.
Compression occurs when a single code is output instead of a string of characters. The
code that the LZW algorithm outputs can be of any arbitrary length, but it must have
more bits in it than a single character. The first 256 codes (when using eight bit characters)
are by default assigned to the standard character set. The remaining codes are assigned
to strings as the algorithm proceeds.

LZW compression works best for files containing lots of repetitive data. This is often the
case with text and monochrome images. Files that are compressed but that do not
contain any repetitive information at all can even grow bigger. LZW compression is fast.
Royalties have to be paid to use LZW compression algorithms within applications (see
below).

LZW compression can be used in a variety of file formats:

• TIFF files
• GIF files

Burrows-Wheeler Transformation (BWT)

The BWT is an algorithm that takes a block of data and rearranges it using a sorting
algorithm. The resulting output block contains exactly the same data elements that it
started with, differing only in their ordering. The transformation is reversible, meaning the
original ordering of the data elements can be restored with no loss of fidelity.

The BWT is performed on an entire block of data at once. Most of today's familiar lossless
compression algorithms operate in streaming mode, reading a single byte or a few bytes
at a time. But with this new transform, there is a need to operate on the largest chunks of
data possible. Since the BWT operates on data in memory, there would be files too big to
process in one fell swoop. In these cases, the file must be split up and processed a block
at a time.
BWT can be used in a variety of file formats

• bzip2
• zip2