Image Compression Using DPCM
Image Compression Using DPCM
ISSN: 2455-5703
Abstract
In this paper we uses The Differential pulse code modulation (DPCM) to remove the unused bit in the image for image
compression. In this paper we compare the compressed image for 1 and 3 bit/pixel using DPCM and also compare the estimation
error. We decrease the compressed image distortion and the estimation error. The PMSE in 3bit/pixel DPCM less 9-10 dB compare
to 1bit/pixel DPCM.
Keywords- Nyquist Sampling, Quantization, Prediction Error, DPCM, PMSE
I. INTRODUCTION
The Differential pulse code modulation (DPCM) [1,2] may be used to remove the unused bit in the image for image compression
In analog messages we can make a good guess about a sample value from knowledge of past sample values. The sample values
are not independent, and generally there is a great deal of redundancy in the Nyquist samples. Proper exploitation of this
redundancy leads to encoding a signal with fewer bits. If x(n) is the nth sample, instead of sending x(n), we send the differentiation
e(n) = x(n) x(n-1). At the output (receiver), knowing e(n) and several earlier sample value x(n-1), we can reconstruct x(n). Thus,
from knowledge of the difference e(n), we can reconstruct x(n) iteratively at the receiver. If this estimation is y(n), then we transmit
the difference (prediction error) e(n)=x(n)-y(n). At the receiver also, we determine the estimated value y(n) from the previous
sample values, and then generate x(n) by adding the received e(n) to the estimate y(n). Thus, we reconstruct the samples at the
receiver iteratively. The predicted (estimated) value y(n) will be close to x(n), and their difference (prediction error) e(n) will be
even smaller than the difference between the successive samples. Consequently, this scheme, known as the differential pulse code
modulation (DPCM) [3], which is a special case of DPCM, where the estimate of a sample value is taken as the previous sample
value, that is, y(n) =x(n-1).
The difference of the original image data (), and prediction image data y(n) is called estimation residual, (). So
() = () () 3.1
is quantized to yield
() = () + () 3.2
Where q(n) is the quantization error, () quantized signal.
() = () () 3.3
I (simg)
() = max = max
3.4
2b 2b
Here b is number of bit. Imax , (simg)max is an image signal.
The prediction output () is fed back to its input so that the predictor input () is
() = () + () 3.5
= () + () 3.6
The prediction input is certainly (), as assumed. The quantized signal () is now transmitted over the channel. Flowchart
diagram of DPCM system is shown below.
histogram plot
800
Oimg
700
600
500
gray level
400
300
200
100
0
0 50 100 150 200 250 300
sample number
IV. SIMULATION RESULT OF IMAGE COMPRESSION FOR 1 AND 3 BIT'S DPCM COMPARISON
MATLAB was used to simulate The DPCM using the application 1bit/pixel and 3bit/pixel image compression comparison. The
256256 novel image is shown in figure 3. This image size is 96.5 kB. This original image pass with 1 and 3 bit/pixel DPCM with
fixed coefficient w [.495 .456] fixed 2taps. Figure 5a shows the reconstructed 1 bit/pixel image. This is a bright image. It is reducing
approximately 13.5kb (13,840bytes) respect to original image for these parameters and the recovered image size 83.0kb
(85,075bytes). Figure 5b shows the reconstructed 3bit/pixel image. This is a dark image. This image was approximately same the
original image. It is reduce approximately 6.4kb (8,672bytes) for these parameters and the recovered image size 88.0kb
(90,243bytes). We can say 1bit/pixel DPCM image reduction was more compare to 3bit/pixel DPCM. But the distortion of 1
bit/pixel DPCM is more compare to 3bit/pixel DPCM. This has been showed in figure 6 and the figure 7 shows histogram plot
comparison between gray label and sample number. The PMSE in 3bit/pixel DPCM less 9-10 dB compare to 1bit/pixel DPCM.
This has been showed in figure 8.
-21.85
-21.95
-22
-22.05
-22.1
-22.15
-22.2
-22.25
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
bit/pixel
Fig. 6: Average square distortion versus transmission bit rate using DPCM
6000
5000
gray level
4000
3000
2000
1000
0
0 50 100 150 200 250 300
sample number
-35
-40
-45
PMSE [dB]
-50
-55
-60
-65
-70
-75
0 50 100 150 200 250 300
sample number
V. CONCLUSION
We can say 1bit/pixel DPCM image reduction was more compare to 3bit/pixel DPCM. But the distortion of 1 bit/pixel DPCM is
more compare to 3bit/pixel DPCM. The PMSE in 3bit/pixel DPCM less 9-10 dB compare to 1bit/pixel DPCM.
REFERENCES
[1] A. Habbi, Comparison of Nth-order DPCM encoder with linear transformation and block quantization techniques, IEEE Trans. Commun., vol. COM-19,
pp. 948-956, Dec. 1971.
[2] R. R. S. Tomar; K. Jain, Lossless Image Compression Using Differential Pulse Code Modulation and its Application IEEE Conference Publications on
Communication Systems and Network Technologies (CSNT), 2015 Fifth International Conference, Pages: 543 545, Year: 2015 DOI:
10.1109/CSNT.2015.192.
[3] B. P. Lathi and Zhi ding Modern Digital and Analog Communication Systems International Fourth Edition. New York Oxford University Press-2010,
pp.292.
[4] J. E. Modestino, and D. G. Daut, Source-channel coding of images, IEEE Trans. Commun., vol. COM-27, pp. 1644-1659, Nov. 1979.
[5] W. K. Pratt, Digital Image Processing, New York: Wiley, 1978.
[6] B. P. Lathi and Zhi ding Modern Digital and Analog Communication Systems International Fourth Edition. New York Oxford University Press-2010,
pp.292