2.4 Use of laplacian pdfs in image compression

Image coding Page 1 / 1

This module introduces the use of Laplacian PDFs in image compression.

It is found to be appropriate and convenient to model the distribution of many types of transformed image coefficients byLaplacian distributions. It is appropriate because much real data is approximately modeled by the Laplacian probabilitydensity function (PDF), and it is convenient because the mathematical form of the Laplacian PDF is simple enough to allowsome useful analytical results to be derived.

A Laplacian PDF is a back-to-back pair of exponential decays and is given by:

p x 1 2 x_{0} x x_{0}

where

x_{0}

is the equivalent of a time constant which defines the width of the PDF from the centre to the

1

points. The initial scaling factor ensures that the area under

p x

is unity, so that it is a valid PDF. shows the shape of

p x

Laplacian PDF,

p x

, and typical quantiser decision thresholds, shown for the case when the quantiser step size

Q 2 x_{0}

The mean of this PDF is zero and the variance is given by:

v x_{0} x

∞ ∞ x 2 p x 2 x x 2 2 x 0 x x 0 2 x 0 2

(using integration by parts twice).

Hence the standard deviation is:

σ x_{0} v x_{0} 2 x_{0}

Given the variance (power) of a subimage of transformed pels, we may calculate

x_{0}

and hence determine the PDF of the subimage, assuming a Laplacian shape. We now show that, if we quantise the subimageusing a uniform quantiser with step size

Q

, we can calculate the entropy of the quantised samples and thus estimate the bit rate needed toencode the subimage in bits/pel. This is a powerful analytical tool as it shows how the compressed bit rate relates directly tothe energy of a subimage. The vertical dashed lines in show the decision thresholds for a typical quantiser for the case when

Q 2 x_{0}

First we analyse the probability of a pel being quantised to each step of the quantiser. This is given by the area under $p x$ between each adjacent pair of quantiser thresholds.

Probability of being at step 0, $p_{0} 12 Q x 12 Q 2 0 x 12 Q$
Probability of being at step $k$ , $p_{k} k 12 Q x k 12 Q$

First, for

x_{2} x_{1} 0

, we calculate:

x_{1} x x_{2} x x_{1} x_{2} p x x_{1} x_{2} 12 x x_{0} 12 x_{1} x_{0} x_{2} x_{0}

Therefore,

p_{0} 1 Q 2 x_{0}

and, for

k 1

p_{k} 12 k 12 Q x_{0} k 12 Q x_{0} Q 2 x_{0} k Q x_{0}

By symmetry, if

k

is nonzero,

p_{- k} p_{k} Q 2 x_{0} k Q x_{0}

Now we can calculate the entropy of the subimage:

H k

∞ ∞ p k 2 logbase --> p k p 0 2 logbase --> p 0 2 k 1 ∞ p k 2 logbase --> p k

To make the evaluation of the summation easier when we substitute for

p_{k}

, we let

p_{k} α r k

where

α Q 2 x_{0}

and

r Q x_{0}

. Therefore,

k 1

∞ p k 2 logbase --> p k k 1 ∞ α r k 2 logbase --> α r k k 1 ∞ α r k 2 logbase --> α k 2 logbase --> r α 2 logbase --> α k 1 ∞ r k α 2 logbase --> r k 1 ∞ k r k

Now

k 1

∞ r k r 1 r and, differentiating by

r

k 1

∞ k r k 1 1 1 r 2 . Therefore,

k 1

∞ p k 2 logbase --> p k α 2 logbase --> α r 1 r α 2 logbase --> r r 1 r 2 α r 1 r 2 logbase --> α 2 logbase --> r 1 r

and

p_{0} 2 logbase --> p_{0} 1 r 2 logbase --> 1 r

Hence the entropy is given by:

H 1 r 2 logbase --> 1 r 2 α r 1 r 2 logbase --> α 2 logbase --> r 1 r

Because both

α

and

r

are functions of

Q x_{0}

, then

H

is a function of just

Q x_{0}

too. We expect that, for constant

Q

, as the energy of the subimage increases, the entropy will also increase approximatelylogarithmically, so we plot

H

against

x_{0} Q

in dB in . This shows that our expectations are born out.

Entropy

H

and approximate entropy

H_{a}

of a quantised subimage with Laplacian PDF, as a function of

x_{0} Q

in dB.

We can show this in theory by considering the case when $≫ x_{0} Q 1$ , when we find that: $α Q 2 x_{0}$ $r 1 Q x_{0} 1 2 α$ $r 1 α$ Using the approximation $2 logbase --> 1 ε ε 2$ for small $ε$ , it is then fairly straightforward to show that $H 2 logbase --> α 12 2 logbase --> 2 x_{0} Q$ We denote this approximation as $H_{a}$ in , which shows how close to $H$ the approximation is, for $x_{0} Q$ (i.e. for $x_{0} Q 0$ dB).

We can compare the entropies calculated using with those that were calculated from the bandpass subimage histograms, as given in these figuresdescribing Haar transform energies and entropies; level 1 energies , level 2 energies , level 3 energies , and level 4 energies . (The Lo-Lo subimages have PDFs which are more uniform and do not fit the Laplacian model well.) The values of $x_{0}$ are calculated from: $x_{0} std. dev. 2 subimage energy 2 (no of pels in subimage)$ The following table shows this comparison:

Transform level	Subimage type	Energy (× $106$ )	No of pels	$x_{0}$	Laplacian entropy	Measured entropy
1	Hi-Lo	4.56	16384	11.80	2.16	1.71
1	Lo-Hi	1.89	16384	7.59	1.58	1.15
1	Hi-Hi	0.82	16384	5.09	1.08	0.80
2	Hi-Lo	7.64	4096	30.54	3.48	3.00
2	Lo-Hi	2.95	4096	18.98	2.81	2.22
2	Hi-Hi	1.42	4096	13.17	2.31	1.75
3	Hi-Lo	13.17	1024	80.19	4.86	4.52
3	Lo-Hi	3.90	1024	43.64	3.99	3.55
3	Hi-Hi	2.49	1024	34.87	3.67	3.05
4	Hi-Lo	15.49	256	173.9	5.98	5.65
4	Lo-Hi	6.46	256	112.3	5.35	4.75
4	Hi-Hi	3.29	256	80.2	4.86	4.38

We see that the entropies calculated from the energy via the Laplacian PDF method (second column from the right) areapproximately 0.5 bit/pel greater than the entropies measured from the Lenna subimage histograms. This is due to the heaviertails of the actual PDFs compared with the Laplacian exponentially decreasing tails. More accurate entropies can beobtained if $x_{0}$ is obtained from the mean absolute values of the pels in each subimage. For a Laplacian PDF we can show that

Mean absolute value x

∞ ∞ x p x 2 x 0 ∞ x 2 x 0 x x 0 x 0

This gives values of

x_{0}

that are about 20% lower than those calculated from the energies and the calculated entropies are then withinapproximately 0.2 bit/pel of the measured entropies.

<< Chapter < Page Page > Chapter >>

Read also:

Get Jobilize Job Search Mobile App in your pocket Now!

100% Free Mobile Applications
Receive real-time job alerts and never miss the right job again

Source: OpenStax, Image coding. OpenStax CNX. Jan 22, 2004 Download for free at http://cnx.org/content/col10206/1.3

Google Play and the Google Play logo are trademarks of Google Inc.

Notification Switch

Would you like to follow the 'Image coding' conversation and receive update notifications?

Ask

©flickr:	Biology 1 By Jill Zerressen Start Quiz
	2 Human A&P 2- Test 2 By Madison Christian Start Flashcards
	3 BOD GI quiz By Brooke Delaney Start Exam
©flickr: Miguel	Gram Positive Infections and Clostridium By Cath Yu Start Quiz
	Biology Final By Anonymous User Start Quiz
	3 Physiotherapy Flashcards Set 3 By Rhodes Start Flashcards
	25 AP 25 Urinary System MCQ By OpenStax Start Quiz
	7 AP 07 Axial Skeleton MCQ By OpenStax Start Quiz
	25 Biology 25 Seedless Plants MCQ By OpenStax Start Quiz
	30 Biology 30 Plant Form and Physiology MCQ By OpenStax Start Quiz