Image and Video Compression

5

2

Lossless Image Compression

2.1 Introduction

Lossless image compression nds applications in medical imaging, indus-

trial applications such as nondestructive testing, satellite imaging and remote

sensing, real-time applications, document management systems, image

archiving, HDTV, and in applications where image quality after decoding

and reconstruction is very good, or no loss of information is expected.

Today, medical images are not stored on lm, making image compres-

sion pivotal. Medical applications use lossless compression instead of lossy

techniques, largely because of legal reasons, even though they provide lower

compression rates compared to lossy compression.

Nondestructive testing (NDT) is a vital part of the quality control process

in the manufacturing industries to detect defects in a manufactured com-

ponent without physically destructing it. Using lossless compression tech-

niques, such challenges can be addressed.

The remotely sensed satellite images have large data volume. This needs to be

signicantly reduced before transmission to earth. The quality of such images is

needed to be as high as possible, because the images are to be used for further sci-

entic analysis. Thus, a technique for image compression that is lossless is required.

Large numbers of applications involve huge data storage and transmission.

In India, we have launched several satellites to collect and send information

to the ground station in connection with the forecasting of weather condi-

tions. These satellites are functioning round the clock, transmitting data to

the ground station periodically. The received data are stored throughout the

day and all days in the year. The data stored for a period of time are analyzed

and used for forecasting the weather. Thus, we need large storage media to

store the data received from the satellite. In practice, employing large size

memory devices involves high cost. Hence, compressing and storing the data

has become a necessity.

Document image compression is a research area that deals with the com-

pression of images of scanned color documents. These documents generally

have a very high resolution and thus are high quality. Such compressed doc-

uments can be transmitted at a high speed even over low-speed connections.

6 Image and Video Compression

It is also possible to have authentic reproduction of the document with

respect to fonts, color, paper texture, and pictures. While compressing, it is

necessary to separate the text and gures from the images. The drawing may

require high spatial resolution and background of a low resolution. So, both

can be coded with a different rate of coding.

A typical lossless compression system consists of similar components as

for lossy technique (i.e., A/D convertor, signal decomposition, quantization,

and lossless coding). A/D converter samples and nely quantizes an image,

producing digital representation in terms of x and y coordinates. Signal

decomposition uses linear transforms. Such decompositions serve to com-

pact the energy into a few coefcients.

The motivation to develop new techniques and stimulation for the rapid

growth of research efforts contribute to the large commercial potential of

image and video coding. It is difcult to achieve a large compression ratio

along with good quality of an image after reconstruction. This also increases

the complexity of the receiver. The characteristics of an ideal image coder

are high delity in the image reconstructed at the receiver, low bit rate, and

reduced complexity of the encoder and the decoder. Input redundancies can

be removed during encoding. The encoded data are transmitted. Decoding

is done at the receiver.

2.2 Source Encoders and Decoders

There is redundancy in an image due to psychovisual or interpixel aspects

and coding. We can use a source encoder to reduce it. The source encoder

should be designed in such a way that it is capable of eliminating cod-

ing, psychovisual, and interpixel redundancies in the input image. In the

source encoder, the rst block is called the mapper, which converts the

given input image into a form that reduces interpixel redundancies. It is

generally a reversible process. The next block is a quantizer block, where the

psychovisual redundancies of the image are minimized. The third block

is the symbol coder that creates a code to assign the quantizer output. The

most frequently occurring values are assigned the shortest code words.

This reduces the coding redundancy. Removing the different types of

redundancies existing in the input data is the main function of the source

encoder.

The source decoder, shown in Figure2.1, contains two blocks. The quan-

tizer block is not present because it is irreversible. The other two blocks have

an inverse operation compared to the encoder.