Dear Kuldip

In paper "PRL 109, 123601 (2012)" in topic "Fundamental Limits of Classical and Quantum Imaging", you can find some good things about your question.

Best

Dear Kuldip,

I think that you need to clarify your question here:  Quantum imaging can mean many things.

I am most familiar with quantum imaging in the context of ghost or entangled imaging.  In such approaches one typically sends one half of an entangled state to an object to be imaged with a bucket detector (low spatial resolution but good temporal resolution), and the other half to an arrayed source.  Correlations between the two detectors therefore allows imaging and indeed some manipulation (e.g. magnification).  As this is a two-qubit protocol, it can be efficiently simulated.

I don't at all understand your point about "simulating Quantum Bits in classical computer may degrade the overall performance": please explain.

Your next point about "image/graphics processing by QBits" is also rather strange: are you looking to find image manipulation algorithms optimised for a quantum computer? Or are you asking about using GPUs to perform simulations of quantum computers?

If you can clarify some of these questions, then we may be able to help.

Dear Sir Andrew D. Greentree,

I mean Quantum imaging in context of Digital Image/Graphics Processing for optimal performance compare to today’s classical computers both hardware and algorithms point of view.

"simulating Quantum Bits in classical computer may degrade the overall performance":

Actually many researcher are doing Image processing by quantum computing approach by using quantum bits. But they have no quantum devices so they are doing their simulation in classical computer by using MATLAB or other software.

In this context one famous professor Joseph Fitzsimons said to me that

“Quantum computing requires information to be stored and manipulated coherently as quantum states, and the current hardware we tend to use for conventional computing simply cannot support these operations. You need a device which can maintain and coherently manipulate quantum states in order to implement a quantum algorithm, and frankly these devices are extremely difficult to build and current attempts only support a handful of qubits (quantum bits...). So, you cannot simply implement a quantum algorithm in matlab, you need new hardware, which to a large extent does not yet exist. You could in principle simulate quantum algorithms classically, but would then loose any performance advantage. It sounds like you are asking for something which does not (and cannot) exist.”

https://www.researchgate.net/post/Has_anyone_ever_done_image_processing_in_MATLAB_by_using_quantum_bits?_tpcectx=profile_highlights

Yes, I am looking to find Image manipulation algorithms optimised for a quantum computer.

Is it possible to prove mathematically entire Image processing task by matrix manipulations (As we know Images can be represented as Matrix) using Quantum Bits to gain more optimal performance?

How can we use Quantum Computing approach for any Image/graphics processing?

Or it’s just an illusion that Quantum Computers are million time faster than classical computer and It’s just made for solving very specific task like factoring of numbers.

Thanks,

Kuldip

OK I see your point now.

I'm no expert on image manipulation, so I don't know enough about the correct way to convert the classical to a quantum algorithm, but there are several points that I may help you.

In reverse order of your points:

"Or it's just an illusion that Quantum Computers are million times faster than classical computer and it's just made for solving very specific task like factoring of numbers"

It is important to realise that classical computation is a subset of quantum computation.  Therefore any task that can be achieved classically can be achieved in a quantum setting, although it is possible that there will be no quantum advantage.  

When comparing fundamental bit vs qubit rates, we typically find that bits are much faster, and additionally in terms of fidelity, bits are more accurate than qubits.  Both of these classical advantages need to be understood and they are not necessarily permanent advantages.

As we currently understand them, bits are encoded in large numbers of electrons in the form of currents, or as magnetic (or optical) domains for memory applications.  This is effectively a form of redundant encoding which permits large signal (and hence digital error correction) and high speed operations.

Conversely, qubits are encoded as a degree of freedom of a fundamental quantum particle.  There is a fundamental mismatch between encoding data and preserving it from noise (systems that are easy to access by an experimenter are typically easy for the environment to access, and hence degrade).  If we could freely copy qubits like we can copy bits, this problem would disappear.

But qubits are more general structures, and the kinds of error correction that are being developed are far more sophisticated.  I conjecture that when we have efficient quantum computers, we will use them for all tasks, not least of all for the fact that the energy per computation will be lower than that available for a classical device.  But we are far from that goal now.

Again, as I said, I'm not an expert in image manipulation, but from what you've suggested, it might be useful for you to consider some of the work that was done in so-called type-II quantum computers.  These were an interesting attempt at interpolating between full-blown QC and small-scale (accessible) QC.  The idea was to create a lattice of classically connected QCs and try to leverage some kind of quantum advantage, or at least test some of the concepts.  Type-II QC has been mainly explored for lattice gas type simulations, but image manipulation where each small-scale QC is responsible for a small part of the image might be another feasible system.  It is also nice, because as Type-II don't have massive entanglement, they are relatively easy to simulate in Matlab (or similar), and you can even make prototypes using FPGAs.  But of course, be aware, that if you can efficiently simulate something, then you are not going to get any kind of dramatic quantum scaleup, and a device that can be simulated with FPGAs can be made with FPGAs just as easily.

I will mention one other idea that I just found, which may be of interest for you to explore (again I don't know if this is an active field or not), but there has been some work on quantum wavelet transformations, which may be of help in studying things like quantum image analysis: see http://arxiv.org/abs/quant-ph/9809004

All the best

Thank you sir for giving me all these valuable information and time.

Regards,

Kuldip

kuldip i hope you will find this document interesting. Read the concluding remarks atleast :)

Unfortunately the exact class of problems a quantum computer can solve is still not very well sketched. In particular, the popular belief that quantum computers will solve all of your problems faster is believed to be false (BQP class is not believed to include NP, although it is believed to be larger than P (take factoring for example) ). It is somewhat funny that we are trying hard to build a quantum computer although we don't really know what it is good for (except factoring). The other applications, like quantum simulation or protein folding, tend to make use of a huge huge number of qubits (since one must use error correction), and it is not yet known whether quantum computation will be of a real use in these cases. Perhaps the most useful today's quantum technology is quantum cryptography, that makes use of nature laws to guarantee complete security (i.e. generating a one-time pad as the cryptographic key), and has already been applied experimentally.

image analysis gives a good results , quantum fields are new it may not give the results you need if not treat it properly.

http://www.iqst.ca/media/pdf/shared/OEI122-CFPs.pdf