Quantum Inspired Algorithms for Image and Signal Processing

Abstract

An Austrian non-university R&D organisation offers novel algorithms that are inspired by concepts from quantum information theory. The algorithms run on conventional computers and are particularly suited to image and signal processing applications involving pattern recognition. Advantages over conventional solutions include the resolution of more signal ambiguities, easier fusion of data from several sources, and the integration of uncertainty and entropy into the encoding of signals.

Description

Quantum Inspired Algorithms (QIA) are novel algorithms that are developed by applying concepts from quantum information to the representation and processing of classical information. That is, they are algorithms that can be implemented and run on conventional computers, whose design is inspired by concepts from quantum information processing, in the same way that evolutionary algorithms are inspired by natural evolution. We have applied this approach to a number of practical problems, and found that it can lead to novel solutions that outperform conventional ones.

The first step in developing a quantum inspired algorithm is to find a suitable encoding of the information as quantum states, which can then be manipulated using the well-developed mathematical techniques of quantum information theory.

In principle quantum inspired algorithms can be developed for any area of information processing, but we have been concentrating on algorithms for image and signal processing that involve pattern recognition. Image and signal data can be represented as probability distributions, which in turn can be represented as quantum states, but these states can be manipulated in ways that are not possible using standard statistical methods, and novel distance measures and correlations between states can be calculated.

A key feature of quantum information is that, in contrast to information represented as probabilities, quantum states can have amplitudes that are negative, allowing them to cancel out as well as to add, which means they can be manipulated in ways that probabilities cannot be.

Innovations and advantages of the offer

  • No signal ambiguities: Signal ambiguities that remain undetected by classical approaches can be detected.
  • Integration of physical quantities: Quantities such as the range of the signals are directly encoded into their quantum representation.
  • Integration of information entropic quantities: Signal parts that contain low entropy have a higher influence on the result than those with high entropy.
  • Integration of uncertainty: Signals that are uncertain with respect to each other are encoded in an appropriate way. Uncertainty becomes a resource rather than a nuisance as the algorithms make a smooth transition from a signal that is becoming more uncertain to one that is more certain.
  • Data fusion: The mathematical quantum space allows for an easier fusion of data from several sources.

Application

Applications for which we have developed quantum inspired algorithms include:

  • An interference-location system for communication satellites
  • A system for curvature-based localization and surveying of urban rail networks
  • A coin recognition and sorting system
  • An optical rail scanner to recognise anomalies in railway track in an early stage, before cracks develop
  • Representation of colour: We have developed a new colour space that shares many properties with human colour perception

Potential applications are in sensor data analysis

  • Oil and gas exploration
  • Medical diagnosis
  • Processing and manufacturing industry

Description of Space Heritage

Space application funded by ARTES 3/4 Phase II (2013) in a contract awarded to Siemens Austria: ONE SATELLITE GEOLOCATION SYSTEM.

Comments on the technology by the broker

The patent owner and his team are eager to adapt their algorithms to challenging tasks and are accustomed to cooperating with industry.

Category
Computer Hardware & Software
Reference No.
TDO0166
Could this technology benefit your business? Please contact Andrea Kurz Brimatech Services GmbH (Austria)
Discussion
comments powered by Disqus