ISSN: 2277–3061 (online) International Journal of Computers & Technology

Volume 2 No.2 April 2012

Image Compression Techniques

H.S Samra
Punjab Polytechnic College
Lalru Mandi, Mohali PB, India

ABSTRACT shown in figure 2.A low resolution 512 ×512×8 bits/pixel × 3

Digital images required large number of bits to represent them color video image require 6×106 bits and over telephone lines of
and in their canonical representation, generally contain 9600 baud (bits/s) modem, the transmission would take
significant amount of redundancy. Image compression approximately 11 minutes for just a single image.
techniques reduce the number of bits required to represent an
image by taking advantage of these redundancies.To overcome
1.3 Why can we compress
this redundancy several image compression techniques are 1.3.1 Spatial Redundancy
discussed in this paper along with their benefits. Neighboring pixels are not independent but correlated. So
unnecessary data repeated within one frame can be removed.
1. INTRODUCTION
1.1Digital Image 1.3.2 Temporal Redundancy
Digital images are composed of pixels and each pixel represents Reduction of the number of bits needed to represent a given
the color at single point in the image. By measuring the color of image or its information.
an image at large number of points, we can create a digital
approximation of the image. Pixels are arranged in regular 1.3.3 Spectral Redundancy
pattern of rows and columns in arrays. A digital image is a This is due to the correlation between different color planes.
numeric representation of a two-dimensional image.
1.2 Image Compression & Its Need
The objective of compression is to reduce redundancy of the
image data in order to enable storing or transmitting data with
minimal space or bandwidth as much as possible, while keeping
the resolution and the visual quality of the reconstructed image
as close to the original image. The principal approach in data
compression is the reduction of the amount of image data (bits)
while preserving information. Image data compression is an
important component of digital camera design and digital
photography.
Figure 2: Digital Image

I/P Mapper Quantizer Encoder 2. BENEFITS OF COMPRESSION

1. The smaller file size that compression provides can take up
Compression much less room on your hard drive, web site or digital camera. It
will also allow for more images to be recorded on other media,
such as a photo CD. Compressed images also take less time to
load than their more cumbersome originals, making it possible
O/P Inverse Dequantizer Decoder to view more images in a shorter period of time.
Mapper 2. On the internet, compressed images not only reduce a web
page's uploading and downloading time. They also take up less
Decompression space on the server in terms of space and bandwidth. so overall
Figure 1: Image Compression & Decompression execution time can be improved.
As shown in figure 1, in first stage, the mapper transforms the 3. It also reduces the probability of transmission errors since
input image into a format designed to reduce inter pixel fewer bits are transferred.
redundancies. The second stage, quantizer block reduces the 4. It also provides a level of security against illicit monitoring.
accuracy of mapper’s output in accordance with a predefined
criterion. In third and final stage, a symbol decoder creates a 3. IMAGE COMPRESSION TECHINQUES
code for quantizer output and maps the output in accordance There are many approaches to image compression, but they can
be categorized into two fundamental groups.
With the code. These blocks perform, in reverse order, the
inverse operations of the encoder’s symbol coder and mapper 1. Lossless Techniques
block .The need of image compression becomes apparent when 2. Lossy Techniques
one computes the number of bits per image resulting from
typical sampling rates. For example, consider the amount of 3.1 Lossless (Reversible) Techniques
storage space and transmission bandwidth required for images as

49 www.ijctonline.com
ISSN: 2277–3061 (online) International Journal of Computers & Technology
Volume 2 No.2 April 2012

Lossless image compression techniques allow the exact original several subintervals, which sizes are proportional to the current
data to be reconstructed from the compressed data. Lossless probability of the corresponding symbols of the alphabet.
compression is used in cases where it is important that the
original and the decompressed data be identical. Compression 3.1.1.2 LZW Encoding
ratio is low. Examples are executable programs, text documents, Lempel Ziv Welch encoding is an example of a category of
and source code. Some image file formats, like PNG or GIF, use algorithms called dictionary-based encoding. The idea is to
only lossless compression. Following techniques (Figure B) are create a dictionary (a table) of strings used during the
included in lossless compression. communication session. If both the sender and the receiver have
a copy of the dictionary, then previously-encountered strings can
be substituted by their index in the dictionary to reduce the
Lossless Compression Techniques amount of information transmitted.
3.1.1.3 Predictive Encoding
Statistical estimation procedure where future random variables
are estimated/predicted from past and present observable
Run Length Statistical Predictive Bit plane random variables. The lossless differential pulse code
Encoding Encoding Encoding Encoding modulation (DPCM) technique is the most common type of
Statistical lossless predictive coding. In the lossless DPCM scheme, each
Figure 3: Taxonomy of Lossless Compression Techniques
Encoding pixel value (except at the boundaries) of the original image is
o
3.1.1Run Length Encoding first predicted based on its neighbors to get a predicted image.
ding
Run-length encoding is probably the simplest method of Then the difference between the actual and the predicted pixel
compression .It is very useful in case of repetitive data. This values is computed to get the differential or residual image. The
technique replaces sequences of identical symbols (pixels), residual image will have a much less dynamic range of pixel
called runs by shorter symbols. The general idea behind this values. This image is then efficiently encoded using Huffman
method is to replace consecutive repeating occurrences of a coding.
symbol by one occurrence of the symbol followed by the 3.1.1.4Bit Plane Encoding
number of occurrences
In most images the neighboring pixels are correlated. That
means the values of the neighboring pixels differ by small
amounts. They can be captured by the representation of pixel
Original data BBBBAAAAANNNNLLLLLL
values in gray code so that the values of neighboring bits in the
Compressed Data bit planes are similar. This makes the individual bit planes. Each
B4A6N4L7 of the bit planes can then be efficiently coded using a lossless
Figure 4: Run Length encoding overview technique. So it is based on decomposing a multilevel image
into a series of binary images and compressing each binary
3.1.1.1Statistical Encoding image.
3.1.1.2Huffman Encoding 3.2 Lossy Techniques
Huffman code is a prefix code. This mean that the code of any
Lossy compression works very differently. These programs
symbol is not prefix of the code of any other symbol. The
principle is to use lower number of bits to encode the data that
occurs more frequently. Huffman coding assigns shorter codes Lossy Compression Techniques
to the symbols that occurs more frequently and longer codes to
that those occurs less frequently. Codes are stored in code book
which may be constructed for each image. Following steps are
used to encode image:
1. Divide image into 8*8 blocks Transform Sub-band Block Truncation
2. Each block is a symbol to be coded. Encoding Encoding Encoding
3. Compute Huffman codes for set of block.
4. Encode block accordingly.
3.1.2.2 Arithmetic Coding
Unlike Huffman coding, arithmetic coding doesn´t use a discrete Vector Fractal Predictive
number of bits for each symbol to compress. the whole data Quantization Encoding Encoding
sequence is coded with a single code. Thus, the correlation
between neighboring pixels is exploited. The biggest drawback Figure 5: Taxonomy of Lossy Compression Techniques
of the arithmetic coding is it´s low speed since of several needed
multiplications and divisions for each symbol. The main idea simply eliminate "unnecessary" bits of information, tailoring the
behind arithmetic coding is to assign to each symbol an interval. file so that it is smaller. This type of compression is used a lot
Starting with the interval [0...1), each interval is divided in for reducing the file size of bitmap pictures, which tend to be
fairly bulky. This may examine the color data for a range of

50 www.ijctonline.com
ISSN: 2277–3061 (online) International Journal of Computers & Technology
Volume 2 No.2 April 2012

pixels, and identify subtle variations in pixel color values that usually the mean of the pixel values in the block. Then a bitmap
are so minute that the human eye/brain is unable to distinguish of the block is derived by replacing all pixels whose values are
the difference between them. The algorithm may choose a greater than or equal (less than) to the threshold by a 1 (0). Then
smaller range of pixels whose color value differences fall within for each segment in the bitmap, the reconstruction value is
the boundaries of our perception, and substitute those for the determined. This is the average of the values of the
others. Lossy compression framework shown in figure 6. corresponding pixels in the original block.

3.2.4 Vector Quantization

It provides a high compression ratio and a simple decoding
Original process. This technique is to develop a dictionary of fixed-size
Image Data Compressed
vectors, called code vectors. A given image is then partitioned
Image
into non-overlapping blocks called image vectors. Then for each
image vector, the closest matching vector in the dictionary is
determined and its index in the dictionary is used as the
Prediction/ Modeling
Quantization encoding of the original image vector. Because of its fast lookup
Transformation/ and
capabilities at the decoder side, VQ-based coding schemes are
Decomposition Encoding
particularly attractive to multimedia applications.
Figure 6: Lossy Compression Framework
3.2.5 Fractal Coding
To achieve this goal one of the following operations is The method is best suited for textures and natural images,
performed. 1. Predicted image is formed by predicting pixels relying on the fact that parts of an image often resemble there
based on values of neighboring pixels of the original image. parts of the same image by converting these parts into
Then residual image is formed which is difference between mathematical data called "fractal codes" which are used to
predicted image and original image. 2. Transformation a recreate the encoded image. Once an image has been converted
reversible process that reduces redundancy and/or provides an into fractal code its relationship to a specific resolution has been
image representation that is more amenable to the efficient lost; it becomes resolution independent. The image can be
extraction and coding of relevant information. 3. Quantization recreated to fill any screen size without the introduction of
process compresses a range of values to a single quantum value. image artifacts or loss of sharpness that occurs in pixel-based
When the number of discrete symbols in a given stream is compression schemes.
reduced, the stream becomes more compressible. Entropy
coding is then applied to achieve further compression. Major 3.2.6 Predictive Encoding
performance considerations of a lossy compression scheme are: The lossy DPCM (Differential Pulse Code Modulation
a) the compression ratio (CR), b) the signal-to noise ratio (SNR) is very similar to the lossless version The major difference is
of the reconstructed image with respect to the original, and c) that in lossy DPCM; the pixels are predicted based on the
the speed of encoding and decoding. “reconstructed values” of certain neighboring pixels. The
difference between the predicted value and the actual value of
3.2.1. Transform Coding the pixels is the differential (residual) image. It is much less
Transform coding algorithm usually start by partitioning the correlated than the original image.
original image into sub images (blocks) of small size (usually 8
x 8). For each block the transform coefficients are calculated, 4. CONCLUSIONS
effectively converting the original 8 x 8 array of pixel values The large use of digital images is expected to continue at an ever
into an array of coefficients closer to the top-left corner usually faster pace in the coming years. The huge size requirements of
contain most of the information needed to quantize and encode images coupled with the explosive increases are straining the
the image with little perceptual distortion. The resulting storage capacities and transmission bandwidths. Compression is
coefficients are then quantized and the output of the quantized a viable way to overcome these bottlenecks. There are basically
isused by a symbol encoding technique to produce the output two types of compression techniques. One is Lossless
bit stream representing the encoded image. Compression and other is Lossy Compression Technique.
3.2.2 Sub-band Coding The best algorithm is measured depends on the following 3
The fundamental concept behind Sub-band Coding (SBC) is that factors:
image is analyzed to produce the components containing
frequencies in well defined bands, the sub bands subsequently, 1. The quality of the image
quantization and coding is applied to each of the bands. The 2. The amount of compression
advantage of this scheme is that the quantization and coding
well suited for each of the sub bands can be designed separately. 3. The speed of compression.

3.2.3 Block Truncation Coding 4. FUTURE SCOPE OF WORK

Block Truncation Coding is a lossy image compression There are good scopes for future improvements in the proposed
techniques .It is a simple technique which involves less compression techniques. In lossless scheme the prediction can
computational complexity. BTC is a recent technique used for be further improved by using Neuro-Fuzzy technique which will
compression of monochrome image data. the image is divided reduce the prediction error further. Image quality and
into non overlapping blocks of pixels. For each block, threshold compression ratio can be improved further by using some
and reconstruction values are determined. The threshold is

51 www.ijctonline.com
ISSN: 2277–3061 (online) International Journal of Computers & Technology
Volume 2 No.2 April 2012

adaptive reversible technique. Computational complexity is

more in most of the algorithms which can be improved further.

5. REFERENCES
[1] Majid Rabanni, Paul W. Jones Volume TT7. Digital Image
Compression Techniques.
[2] Subramanya A, “Image Compression Techniques
“.Potentials IEEE, Vol.20, Issue 1, Feb-March 2001.
[3] Ming Yang & Nikolaos Bourbakis ,“An Overview Of
Lossless Digital Image Compression Techniques,” Circuits
& Systems, 2005 48th Midwest Symposium vol. 2 IEEE ,7
– 10 Aug, 2005.
[4] Doaa Mohammed ,Fatma Abou-Chadi. “Image
Compression using Block Truncation Coding”. JSAT
February edition 2011.
[5] David H. Kil and Fances Bongjoo Shin, “ Reduced
Dimension Image Compression And its
applications,”Image Processing, 1995, Proceedings,
International Conference,Vol. 3 , Oct.,1995.
[6] C.K. Li and H.Yuen, “A High Performance Image
Compression Technique For Multimedia Applications,”
IEEE Transactions on Consumer Electronics, Vol. 42,
no. 2, 2 May 1996.
[7] Sindhu M, Raj Kamal R , “ Images and Its Compression
Techniques” , International Journal of Recent Trends in
Engineering , Vol 2,No.4, November 2009.
[8] K. Sayood, Introduction to Data Compression (2nd
edition), Morgan-Kaufmann, 2000.

52 www.ijctonline.com