Internal representation of a digital image inside a quantum processor

Framework of Quantum Image Processing
Spread the love

Scientists at Florida International University have compared five techniques for the representation of a digital image inside a quantum processor.

Since its inception nearly two decades ago, Quantum Image Processing (QImP) has always dealt with the same problem, i.e., the internal representation of a digital image inside a quantum circuit efficiently, where such circuits can be optical or of superconductors. In the case of superconducting quantum platforms, they have been freely available to the entire scientific community for approximately five years, which has allowed testing the different techniques for the internal representation of an image on a real physical machine without the need for theoretical speculations. However, during the last five years we have witnessed a complete absence of such implementations.

From all the accumulated experience in Quantum Information Processing, the scientific community knows that the problem with simulators is that they represent a necessary but not sufficient condition, i.e., if something works in a simulator, e.g. Qiskit, it still needs to be tested on a QPU, but if something does not work in a simulator, then do not even bother to move to the QPU because it is clear that our quantum algorithm under test has problems.

The techniques are: flexible representation of quantum images (FRQI), novel enhanced quantum representation (NEQR), generalized quantum image representation (GQIR), multi-channel representation for quantum images (MCQI), and quantum Boolean image processing (QBIP).

The comparison has been based on implementations on the Quirk simulator, and on the IBM Q Experience processors, from the point of view of performance, robustness (noise immunity), deterioration of the outcomes due to decoherence, and technical viability.

The paper can be read there.