1
LOW BIT-RATE IMAGE COMPRESSION VIA ADAPTIVE DOWN
SAMPLING AND CONSTRAINED LEAST SQUARES UPCONVERSION
R. Ashok Kumar Reddy, Project Associate, ashokreddy02@gmail.com
Abstract
In this paper, we are going to use a practical
approach of uniform down sampling in image space and
yet making the sampling adaptive by spatially varying,
directional low-pass pre-filtering. The resulting down-
sampled pre-filtered image remains a conventi onal square
conventional
sample grid, and, thus, it can be compressed and
transmitted without any change to current image coding
standards and systems. The decoder first decompresses
the low-resolution image and then up-converts it to the
original resolution in a cons trained least squares
constrained
restoration process, using a 2-D piecewise autoregressive
model and the knowledge of directional low-pass pre-
filtering. The proposed compression approach of
collaborative adaptive down-sampling and up-conversion
(CADU) outperforms JPEG 2000 in PSNR measure at low
to medium bit rates and achieves superior visual quality,
as well. The superior low bit-rate performance of the
CADU approach seems to suggest that over-sampling not
only wastes hardware resources and energy, and it could
be counterproductive to image quality given a tight bit
budget.
1. Introduction:
The term digital image refers to processing of a
two dimensional picture by a digital computer. In a
broader context, it implies digital processing of any two
dimensional data. A digital image is an array of real or
complex numbers represented by a finite number of bits.
An image given in the form of a transparency, slide,
photograph or an X-rays is first digitized and stored as a
matrix of binary digits in computer memory. This
digitized image can then be processed and or displayed
on a high-resolution television monitor. For display, the
image is stored in a rapid-access buffer memory, which
refreshes the monitor at a rate of 25 frames per second
to produce a visually continuous display.
1.1 The Image Processing System:
A typical digital image processing system is
given in fig.1.1
1.1.1 Digitizer
A digitizer converts an image into a
numerical representation suitable for input into a
digital computer. Some common digitizers are
1. Microdensitometer
2. Flying spot scanner
3. Image dissector
4. Videocon camera
5. Photosensitive solid- state arrays.
1.1.2 Image Processor
An image processor does the functions of
image acquisition, storage, preprocessing,
segmentation, representation, recognition and
interpretation and finally displays or records the
resulting image. The following block diagram
gives the fundamental sequence involved in an
image processing system.
As detailed in the diagram, the first step in
the process is image acquisition by an imaging
sensor in conjunction with a digitizer to digitize the
image. The next step is the preprocessing step
where the image is improved being fed as an input
to the other processes. Preprocessing typically
deals with enhancing, removing noise, isolating
regions, etc. Segmentation partitions an image into
its constituent parts or objects. The output of
segmentation is usually raw pixel data, which
consists of either the boundary of the region or the
pixels in the region themselves. Representation is
the process of transforming the raw pixel data into
a form useful for subsequent processing by the
computer. Description deals with extracting
features that are basic in differentiating one class of
objects from another. Recognition assigns a label
to an object based on the information provided by
its descriptors. Interpretation involves assigning
meaning to an ensemble of recognized objects. The
knowledge about a problem domain is incorporated
into the knowledge base. The knowledge base
guides the operation of each processing module
and also controls the interaction between the
modules. Not all modules need be necessarily
present for a specific function. The composition of
the image processing system depends on its

application. The frame rate of the image processor is
normally around 25 frames per second.
1.1.3 Digital Computer:
Mathematical processing of the digitized image
such as convolution, averaging, addition, subtraction,
etc. are done by the computer.
1.1.4 Mass Storage
The secondary storage devices normally used are
floppy disks, CD ROMs etc.
1.1.5 Hard Copy Device
The hard copy device is used to produce a
permanent copy of the image and for the storage of the
software involved.
1.1.6 Operator Console
The operator console consists of equipment
and arrangements for verification of intermediate
results and for alterations in the software as and when
require. The operator is also capable of checking for
any resulting errors and for the entry of requisite data.
2. Image processing fundamentals:
2.1 Introduction
Digital image processing refers processing of
the image in digital form. Modern cameras may directly
take the image in digital form but generally images are
originated in optical form. They are captured by video
cameras and digitalized. The digitalization process
includes sampling, quantization. Then these images are
processed by the five fundamental processes, at least
any one of them, not necessarily all of them.
2.2 Image Processing Techniques:
This section gives various image processing
techniques.
Fig 2.2.1 Image processing Techniques
2
2.2.1 Image Enhancement
Image enhancement operations improve the
qualities of an image like improving the image’s
contrast and brightness characteristics, reducing its
noise content, or sharpen the details. This just
enhances the image and reveals the same
information in more understandable image. It does
not add any information to it.
2.2.2 Image Restoration
Image restoration like enhancement improves
the qualities of image but all the operations are
mainly based on known, measured, or degradations
of the original image. Image restorations are used
to restore images with problems such as geometric
distortion, improper focus, repetitive noise, and
camera motion. It is used to correct images for
known degradations.
2.2.3 Image Analysis
Image analysis operations produce numerical
or graphical information based on characteristics of
the original image. They break into objects and
then classify them. They depend on the image
statistics. Common operations are extraction and
description of scene and image features, automated
measurements, and object classification. Image
analyze are mainly used in machine vision
applications.
2.2.4 Image Compression
Image compression and decompression reduce
the data content necessary to describe the image.
Most of the images contain lot of redundant
information, compression removes all the
redundancies. Because of the compression the size
is reduced, so efficiently stored or transported. The
compressed image is decompressed when
displayed. Lossless compression preserves the
exact data in the original image, but Lossy
compression does not represent the original image
but provide excellent compression.
2.2.5 Image Synthesis
Image synthesis operations create images from
other images or non-image data. Image synthesis
operations generally create images that are either
physically impossible or impractical to acquire.
2.3 Applications
Digital image processing has a broad spectrum
of applications, such as remote sensing via
satellites and other spacecrafts, image transmission
and storage for business applications, medical
processing, radar, sonar and acoustic image
processing, robotics and automated inspection of
industrial parts.

2.3.1 Medical Applications
In medical applications, one is concerned with
processing of chest X-rays, cineangiograms, projection
images of transaxial tomography and other medical
images that occur in radiology, nuclear magnetic
resonance (NMR) and ultrasonic scanning. These
images may be used for patient screening and
monitoring or for detection of tumors or other disease
in patients.
2.3.2 Satellite Imaging
Images acquired by satellites are useful in tracking
of earth resources; geographical mapping; prediction of
agricultural crops, urban growth and weather; flood and
fire control; and many other environmental applications.
Space image applications include recognition and
analysis of objects contained in image obtained from
deep space-probe missions.
2.3.3 Communication
Image transmission and storage applications occur
in broadcast television, teleconferencing, and
transmission of facsimile images for office automation,
communication of computer networks, closed-circuit
television based security monitoring systems and in
military communications..
2.3.4 Radar Imaging Systems
Radar and sonar images are used for detection and
recognition of various types of targets or in guidance
and manoeuvring of aircraft or missile systems.
2.3.5 Document Processing
It is used in scanning, and transmission for
converting paper documents to a digital image form,
compressing the image, and storing it on magnetic tape.
It is also used in document reading for automatically
detecting and recognizing printed characteristics.
2.3.6 Defense/Intelligence
It is used in reconnaissance photo-interpretation for
automatic interpretation of earth satellite imagery to
look for sensitive targets or military threats and target
acquisition and guidance for recognizing and tracking
targets in real-time smart-bomb and missile-guidance
systems.
Need for compression:
• To reduce the volume of data to be transmitted
(text, fax, images)
• To reduce the bandwidth required for transmission
and to reduce storage requirements (speech, audio,
video)
• How is compression possible?
• Redundancy in digital audio, image, and video data
• Properties of human perception
3
• Digital audio is a series of sample values;
image is a rectangular array of pixel values;
video is a sequence of images played out at a
certain rate
• Neighboring sample values are correlated
• Transmission and storage of uncompressed
video would be extremely costly and
impractical.
• Frame with 352x288 contains 202,752 bytes of
information.
• Recoding of uncompressed version of this
video at 15 frames per second would require 3
MB. One minute�180 MB storage. One 24-
hour day�262 GB
• Using compression, 15 frames/second for 24
hour�1.4 GB, 187 days of video could be
stored using the same disk space that
uncompressed video would use in one day
Redundancy
• Adjacent audio samples are similar (predictive
encoding); samples corresponding to silence
(silence removal)
• In digital image, neighboring samples on a
scanning line are normally similar (spatial
redundancy)
• In digital video, in addition to spatial
redundancy, neighboring images in a video
sequence may be similar (temporal
redundancy)
Principles of Compression
Spatial Correlation:
• Redundancy among neighboring pixels
Spectral Correlation:
• Redundancy among different color planes
Temporal Correlation:
• Redundancy between adjacent frames in a
sequence of image
Human Perception Factors
• Compressed version of digital audio, image,
video need not represent the original information
exactly
• Perception sensitivities are different for
different signal patterns.
• Human eye is less sensitive to the higher
spatial frequency components than the lower
frequencies (transform coding)
Classification
• Lossless compression
– lossless compression for legal and medical
documents, computer programs
– Exploit only data redundancy.

• Lossy compression
– digital audio, image, video where some errors or
loss can be tolerated
– Exploit both data redundancy and human
perception properties.
Predictive coding:
• Information already received (in transmission) is
used to predict future values
• Difference between predicted and actual is stored.
• Easily implemented in spatial (image) domain
• Example: Differential Pulse Code Modulation
(DPCM).
Transform Coding:
• Transform signal from spatial domain to other
space using a well-known transform
• Encode signal in new domain (by string
coefficients)
• Higher compression, in general than predictive, but
requires more computation (apply quantization).
Sub band Coding:
• Split the frequency band of a signal in various
subbands.
• The filters used in subband coding are known as
quadrature mirror filter(QMF)
• Use octave tree decomposition of an image data
into various frequency subbands.
• The output of each decimated subbands quantized
and encoded separately
• Constant bit rate versus variable bit rate coding
Block diagram:
Fig. 1. Block diagram of the proposed CADU
image compression system.
Existing system:
Compressive Sampling: This paper surveys an
emerging theory which goes by the name of
“compressive sampling” or “compressed sensing,” and
which says that this conventional wisdom is inaccurate.
Perhaps surprisingly, it is possible to reconstruct images
or signals of scientific interest accurately and
sometimes even exactly from a number of samples
which is far smaller than the desired resolution of the
image/signal, e.g. the number of pixels in the image.
4
Disadvantages:
• The number of Fourier samples we need to
acquire must match the desired resolution of the
image, i.e. the number of pixels in the image.
Proposed System
We propose a new, standard-compliant
approach of coding uniformly down-sampled
images, which outperforms JPEG 2000 in both
PSNR and visual quality at low to modest bit rates.
This success is due to the novel up-conversion
process of least square non-causal predictive
decoding, constrained by adaptive directional low-
pass pre-filtering. Our findings suggest that a lower
sampling rate can actually produce higher quality
images at certain bit rates. By feeding the standard
methods down-sampled images, the new approach
reduces the workload and energy consumption of
the encoders, which is important for wireless visual
communication.
Advantages
• The proposed compression approach of
collaborative adaptive down-sampling and up-
conversion (CADU) outperforms JPEG 2000 in
PSNR measure at low to medium bit rates and
achieves superior visual quality, as well.
Domain: Digital Image Processing
Digital Image processing is the use of
computer algorithms to perform image processing
on Digital Images. As a subfield of digital signal
processing, digital image processing has many
advantages over analog image processing; it allows
a much wider range of algorithms to be applied to
the input data, and can avoid problems such as the
build-up of noise and signal distortion during
processing.
Software Requirement
MATLAB 7.0 and above MATLAB is a high-
performance language for technical computing. It
integrates computation, visualization, and
programming in an easy-to-use environment where
problems and solutions are expressed in familiar
mathematical notation. Typical uses include:
• Math and computation Algorithm development
• Modeling, simulation, and prototyping
• Data analysis, exploration, and visualization
• Scientific and engineering graphics
• Application development, including Graphical
User Interface building.
MATLAB is an interactive system whose
basic data element is an array that does not require
dimensioning. This allows you to solve many
technical computing problems, especially those
with matrix and vector formulations, in a fraction

of the time it would take to write a program in a scalar
non-interactive language such as C or FORTRAN.
Introduction:
The prevailing engineering practice of image/video
compression usually starts with a dense 2-D sample
grid of pixels. Compression is done by transforming the
spatial image signal into a space (e.g., spaces of Fourier
or wavelet bases) in which the image has a sparse
representation and by entropy coding of transform
coefficients. Recently, researchers in the emerging field
of compressive sensing introduced a new method called
“over sampling followed massive dumping” approach.
They showed, quite surprisingly, it is possible, at least
theoretically, to obtain compact signal representation by
a greatly reduced number of random samples
This project investigates the problem of compact
image representation in an approach of sparse sampling
in the spatial domain. The fact that most natural images
have an exponentially decaying power spectrum
suggests the possibility of interpolation-based compact
representation of images. A typical scene contains
predominantly smooth regions that can be satisfactorily
interpolated from a sparsely sampled low-resolution
image. The difficulty is with the reconstruction of high
frequency contents. Of particular importance is faithful
reconstruction of edges without large phase errors,
which is detrimental to perceptual quality of a decoded
image. For all these drawbacks, new image
compression methodology of collaborative adaptive
down-sampling and up conversion (CADU).
Scope Of The Project:
The main objective is to propose a new,
standard-compliant approach of coding uniformly
down-sampled images, which outperforms JPEG 2000
in both PSNR and visual quality at low to modest bit
rates by using the novel up conversion process of least
square noncausal predictive decoding, constrained by
adaptive directional low-pass prefiltering. It is to
estimate that a lower sampling rate can actually produce
higher quality images at certain bit rates.
Module:
Module 1: Decomposition of low –resolution image.
Module 2: Upconversion of the image to its resolution
by PAR .
Module 3: Reverse the directional low-pass prefiltering
operation of the encoder.
Module Description:
The CADU decoder first decompresses the low-
resolution image and then upconverts it to the original
resolution in constrained least squares restoration
process, using a 2-D piecewise autoregressive model
(PAR) and by reversing the directional low-pass
prefiltering operation of the encoder. Two-dimensional
autoregressive modeling was a known effective
5
technique of predictive image coding For the
CADU decoder; the PAR model plays a role of
adaptive noncausal predictor. The CADU approach
is very novel and unique that the predictor is only
used at the decoder side, and the noncausal
predictive decoding is performed in collaboration
with the prefiltering of the encoder.
Module 1:
The CADU image compression technique,
although operating on down-sampled images,
obtains some of the best PSNR results and visual
quality at low to medium bit rates. CADU
outperforms the JPEG 2000 standard, even though
the latter is fed images of higher resolution and is
widely regarded as an excellent low bit-rate image
codec. Since the down-sampled image has the
conventional form of square pixel grid and can be
fed directly to any existing image codec, standard
or proprietary, the CADU up conversion process is
entirely up to the decoder the proposed CADU
image coding approach can work in tandem with
any third party image/video compression
techniques. This flexibility makes standard
compliance a non issue for the new CADU method.
We envision that CADU becomes a useful
enhancer of any existing image compression
standard for improved low bit-rate performance.
We make a more compact representation of an
image by decimating every other row and every
other column of the image. This simple approach
has an operational advantage that the down-
sampled image remains a uniform rectilinear grid
of pixels and can readily be compressed by any of
existing international image coding standards. To
prevent the down-sampling process from causing
aliasing artifacts, it seems necessary to low-pass
prefilter an input image to half of its maximum
frequency. However, on a second reflection, one
can do somewhat better. In areas of edges, the 2-D
spectrum of the local image signal is not isotropic.
Thus, we seek to perform adaptive sampling,
within the uniform down-sampling
framework, by judiciously smoothing the image
with directional low-pass prefiltering prior to
down-sampling.
To this end, we design a family of 2-D
directional low-pass prefilters under the criterion of
preserving the maximum 2-D bandwidth without
the risk of aliasing. Let WL(θ) and WH(θ) be the
side lengths of the rectangular low-passed region of
the 2-D filter in the low- and high-frequency
directions of an edge of angle , respectively. The
maximum area of this low-passed region without
aliasing is
A= WL (θ) .WH (θ) = π2
 6

as that between its reference and its neighbors in

It is easy to show that there are only eight values of the low resolution version. Therefore, the PAR
(corresponding to three combinations of and values) to model parameters can be estimated from the
achieve while avoiding aliasing. These eight cases are decoded image by solving the following two least-
tabulated in Table I. Fig. 2 illustrates the above square estimation problems.
directional low-pass filter design for , (the lowpassed Under the assumption of piecewise stationarity
frequency range for other angles can be obtained by of the image signal, the decoder carries out the
rotation). The spectra in the diagrams are those of the proposed up sampling process on a block of pixels
straight line of angle .In addition, the directional low- at a time. To prevent possible block visual artifacts,
pass filter design serves two other purposes: 1) most the neighboring blocks should have some overlaps.
efficient packing of signal energy in presence of edges; The choice of block size and the degree of spatial
2) preservation of subjective image quality for the edge overlap are design details related to
is an important semantic construct. Moreover, as we implementation complexity and viewer preference
will see in the next section, the use of low-pass on image attributes. The CADU system design is
prefilters establishes sample relations that play a central asymmetric: the encoder is a simple and
role in the decoding process of constrained least squares inexpensive process, while the decoder involves
upconversion. Many implementations of directional solving a rather large-scale optimization problem
low-pass prefilters are possible. In the directional described . The computation bottleneck is in
prefiltering step, the CADU encoder first computes the inverting an matrix in , where is the number of
gradient at the sampled position. pixels to be jointly recovered. Instead of inverting
If the amplitude of the gradient is below a the matrix directly, we solve numerically via
threshold, the isotropic low-pass filter is applied. differentiation using the conjugate gradient method.
Otherwise, the gradient direction is quantized and the The solution is guarantied to be globally optimal
corresponding filter is selected and applied. Despite its for the objective function is convex.
simplicity, the CADU compression approach via
uniform down-sampling is not inherently inferior to Results:
other image compression techniques in rate-distortion
performance, as long as the target bit rate is below a
threshold.
Module 2:
In the directional prefiltering step, the CADU
encoder first computes the gradient at the sampled
position. If the amplitude of the gradient is below a
threshold, the isotropic low-pass filter is applied.
Otherwise, the gradient direction is quantized and the
corresponding filter is selected and applied. Despite its
simplicity, the CADU compression approach via
uniform down-sampling is not inherently inferior to
other image compression techniques in rate-distortion
performance, as long as the target bit rate is below a
threshold. The argument is based on the classical water-
filling principle in rate-distortion theory. To encode a
set of independent Gaussian random variables, the rate-
distortion bounds, the total bit rate and the total mean-
squares distortion are calculated.
Most natural images have a rapidly (e.g.,
exponentially) decaying power spectrum . Suppose that
the input image is in the Fourier domain then its power
spectrum is monotonically decreasing.
Module 3:
Then the next step is to estimate the model CONCLUSION:
parameters. Recall that our design aims to reconstruct We proposed a new, standard-compliant
significant edges. Because edges have high correlation approach of coding uniformly down-sampled
in the scale space, the directional correlation of that images, which outperforms JPEG 2000 in both
pixel and its neighbors in the original image is the same PSNR and visual quality at low to modest bit
 7

rates. This success is due to the novel up conversion

process of least square noncausal predictive decoding,
constrained by adaptive directional low-pass
prefiltering. Our findings suggest that a lower sampling
rate can actually produce higher quality images at
certain bit rates. By feeding the standard methods down
sampled images, the new approach reduces the
workload and energy consumption of the encoders,
which is important for wireless visual communication.

REFERENCES:
[1] E. Cands, “Compressive sampling,” in Proc. Int.
Congr. Mathematics, Madrid, Spain, 2006, pp.
1433–1452. WU et al.: LOW BIT-RATE IMAGE
COMPRESSION VIA ADAPTIVE DOWN-
SAMPLING 561
[2] X.Wu,K.U. Barthel, and W. Zhang, “Piecewise 2-D
autoregression for predictive image coding,” in Proc.
IEEE Int. Conf. Image Processing, Chicago, IL, Oct.
1998, vol. 3, pp. 901–904.
[3] X. Li and M. T. Orchard, “Edge-direted prediction
for lossless compression of natural images,” IEEE
Trans. Image Process., vol. 10, no. 6, pp. 813–817, Jun.
2001.
[4] D. Santa-Cruz, R. Grosbois, and T. Ebrahimi, “Jpeg
2000 performance evaluation and assessment,” Signal
Process.: Image Commun., vol. 1, no. 17, pp. 113–130,
2002.
[5] A. M. Bruckstein, M. Elad, and R. Kimmel, “Down-
scaling for better transform compression,” IEEE Trans.
Image Process., vol. 12, no. 9, pp. 1132–1144, Sep.
2003.
[6] Y. Tsaig, M. Elad, and P. Milanfar, “Variable
projection for near-optimal filtering in low bit-rate
block coders,” IEEE Trans. Circuits Syst. Video
Technol., vol. 15, no. 1, pp. 154–160, Jan. 2005.
[7] W. Lin and D. Li, “Adaptive downsampling to
improve image compression at low bit rates,” IEEE
Trans. Image Process., vol. 15, no. 9, pp. 2513–2521,
Sep. 2006.
[8] L. Gan, C. Tu, J. Liang, T. D. Tran, and K.-K. Ma,
“Undersampled boundary pre-/post-filters for low bit-
rate dct-based coders,” IEEE Trans. Image Process. ,
vol. 16, no. 2, pp. 428–441, Feb. 2007.
[9] X. Zhang, X. Wu, and F. Wu, “Image coding on
quincunx lattice with adaptive lifting and
interpolation,” in Proc. IEEE Data Compression Conf.,
Mar. 2007, pp. 193–202.
[10] B. Zeng and A. N.Venetsanopoulos, “A jpeg-based
interpolative image coding scheme,” in Proc. IEEE
ICASSP, 1993, vol. 5, pp. 393–396.
[11] D. Tabuman and M. Marcellin , JPEG2000: Image
Compression Fundamentals, Standards and Parctice.
Norwell, MA: Kluwer, 2002.