<< Chapter < Page Chapter >> Page >
This collection reviews fundamental concepts underlying the use of concise models for signal processing. Topics are presented from a geometric perspective and include low-dimensional linear, sparse, and manifold-based signal models, approximation, compression, dimensionality reduction, and Compressed Sensing.

Transform coding

In Nonlinear Approximation from Approximation , we measured the quality of a dictionary in terms of its K -term approximations to signals drawn from some class. One reason that such approximations are desirableis that they provide concise descriptions of the signal that can be easily stored, processed, etc. There is even speculation andevidence that neurons in the human visual system may use sparse coding to represent a scene [link] .

For data compression, conciseness is often exploited in a popular technique known as transform coding . Given a signal f (for which a concise description may not be readily apparent in itsnative domain), the idea is simply to use the dictionary Ψ to transform f to its coefficients α , which can then be efficiently and easily described. As discussed above, perhaps thesimplest strategy for summarizing a sparse α is simply to threshold, keeping the K largest coefficients and discarding the rest. A simple encoder would then just encode thepositions and quantized values of these K coefficients.

Metric entropy

Suppose f is a function and let f R ^ be an approximation to f encoded using R bits. To evaluate the quality of a coding strategy, it is common to consider the asymptotic rate-distortion (R-D) performance, which measures the decay rate of f - f R ^ L p as R . The metric entropy   [link] for a class F gives the best decay rate that can be achieved uniformly over all functions f F . We note that this is a true measure for the complexity of a class and is tied to noparticular dictionary or encoding strategy. The metric entropy also has a very geometric interpretation, as it relates to thesmallest radius possible for a covering of 2 R balls over the set F .

Metric entropies are known for certain signal classes. For example, the results of Clements  [link] (extending those of Kolmogorov and Tihomirov  [link] ) regarding metric entropy give bounds on the optimal achievable asymptoticrate-distortion performance for D -dimensional C H -smooth functions f (see also [link] ):

f - f R ^ L p 1 R H D .
Rate-distortion performance measures the complexity of a representation and encoding strategy. In the case of transformcoding, for example, R-D results account for the bits required to encode both the values of the significant coefficients and their locations. Nonetheless, in many cases transform coding is indeed an effective strategy for encoding signals that have sparserepresentations  [link] . For example, in  [link] Cohen et al. propose a wavelet-domain coder that uses a connected-tree structure to efficiently encode the positions ofthe significant coefficients and prove that this encoding strategy achieves the optimal rate
f - f R ^ L p 1 R H D .

Compression of piecewise smooth images

In some cases, however, the sparsity of the wavelet transform may not reflect the true underlying structure of a signal. Examplesare 2-D piecewise smooth signals with a smooth edge discontinuity separating the smooth regions. As we discussed in Nonlinear Approximation from Approximation , wavelets fail to sparsely represent these functions, and so the R-D performance for simplethresholding-based coders will suffer as well. In spite of all of the benefits of wavelet representations for signal processing (lowcomputational complexity, tree structure, sparse approximations for smooth signals), this failure to efficiently represent edgesis a significant drawback. In many images, edges carry some of the most prominent and important information [link] , and so it is desirable to have a representation well-suited tocompressing edges in images.

To address this concern, recent work in harmonic analysis has focused on developing representations that provide sparsedecompositions for certain geometric image classes. Examples include curvelets  [link] , [link] and contourlets  [link] , slightly redundant tight frames consisting of anisotropic, “needle-like” atoms.In  [link] , bandelets are formed by warping an orthonormal wavelet basis to conform to the geometrical structurein the image. A nonlinear multiscale transform that adapts to discontinuities (and can represent a “clean” edge using very fewcoarse scale coefficients) is proposed in  [link] . Each of these new representations has been shown to achievenear-optimal asymptotic approximation and R-D performance for piecewise smooth images consisting of C H regions separated by discontinuities along C H curves, with H = 2 ( H 2 for bandelets). Some have also found use in specialized compression applications such asidentification photos  [link] .

In [link] , we have presented a scheme that is based on the simple yet powerful observation that geometric features can be efficientlyapproximated using local, geometric atoms in the spatial domain, and that the projection of these geometric primitives onto waveletsubspaces can therefore approximate the corresponding wavelet coefficients. We prove that the resulting dictionary achieves theoptimal nonlinear approximation rates for piecewise smooth signal classes. To account for the added complexity of this encodingstrategy, we also consider R-D results and prove that this scheme comes within a logarithmic factor of the optimal performance rate.Unlike the techniques mentioned above, our method also generalizes to arbitrary orders of smoothness and arbitrary signal dimension.

Questions & Answers

what does nano mean?
Anassong Reply
nano basically means 10^(-9). nanometer is a unit to measure length.
do you think it's worthwhile in the long term to study the effects and possibilities of nanotechnology on viral treatment?
Damian Reply
absolutely yes
how to know photocatalytic properties of tio2 nanoparticles...what to do now
Akash Reply
it is a goid question and i want to know the answer as well
characteristics of micro business
for teaching engĺish at school how nano technology help us
Do somebody tell me a best nano engineering book for beginners?
s. Reply
what is fullerene does it is used to make bukky balls
Devang Reply
are you nano engineer ?
fullerene is a bucky ball aka Carbon 60 molecule. It was name by the architect Fuller. He design the geodesic dome. it resembles a soccer ball.
what is the actual application of fullerenes nowadays?
That is a great question Damian. best way to answer that question is to Google it. there are hundreds of applications for buck minister fullerenes, from medical to aerospace. you can also find plenty of research papers that will give you great detail on the potential applications of fullerenes.
what is the Synthesis, properties,and applications of carbon nano chemistry
Abhijith Reply
Mostly, they use nano carbon for electronics and for materials to be strengthened.
is Bucky paper clear?
so some one know about replacing silicon atom with phosphorous in semiconductors device?
s. Reply
Yeah, it is a pain to say the least. You basically have to heat the substarte up to around 1000 degrees celcius then pass phosphene gas over top of it, which is explosive and toxic by the way, under very low pressure.
Do you know which machine is used to that process?
how to fabricate graphene ink ?
for screen printed electrodes ?
What is lattice structure?
s. Reply
of graphene you mean?
or in general
in general
Graphene has a hexagonal structure
On having this app for quite a bit time, Haven't realised there's a chat room in it.
what is biological synthesis of nanoparticles
Sanket Reply
what's the easiest and fastest way to the synthesize AgNP?
Damian Reply
types of nano material
abeetha Reply
I start with an easy one. carbon nanotubes woven into a long filament like a string
many many of nanotubes
what is the k.e before it land
what is the function of carbon nanotubes?
I'm interested in nanotube
what is nanomaterials​ and their applications of sensors.
Ramkumar Reply
what is nano technology
Sravani Reply
what is system testing?
preparation of nanomaterial
Victor Reply
how did you get the value of 2000N.What calculations are needed to arrive at it
Smarajit Reply
Privacy Information Security Software Version 1.1a
Berger describes sociologists as concerned with
Mueller Reply
Got questions? Join the online conversation and get instant answers!
QuizOver.com Reply

Get the best Algebra and trigonometry course in your pocket!

Source:  OpenStax, Concise signal models. OpenStax CNX. Sep 14, 2009 Download for free at http://cnx.org/content/col10635/1.4
Google Play and the Google Play logo are trademarks of Google Inc.

Notification Switch

Would you like to follow the 'Concise signal models' conversation and receive update notifications?