This thesis examines the application of Digital Signal Processing (DSP) techniques, and specifically Wavelets, to the field of automatic coin recognition. The aim is to utilise DSP techniques to exploit information that is contained within time domain signals representing coins, which can not be accessed by other means. Attention is also given to the power requirement of possible solutions, with a low power solution being a secondary aim, as the solutions are targeted for use in a line-powered payphone. An examination of existing coin recognition techniques is presented, for which an improved but basic DSP coin recognition scheme using peak and trough location, is achieved. This is then improved using more advanced DSP techniques to access previously unavailable information contained within the signals. The advanced DSP techniques are developed into an integrated framework for automatic coin recognition. The framework is used to identify a single Wavelet solution that supplies a DSP representation of a set of coins. The representations of different coin types exist within a region of n-dimensional Euclidean space, which the framework attempts to locate uniquely for each coin type. To enable the framework to operate successfully, a key feature presented is the resampling of the waveforms input into the framework, to normalise any temporal variations in the input data. The location of the single Wavelet for analysis can not be achieved analytically and so is obtained using a novel Data Mining solution to search a Wavelet dictionary for possible solutions. This thesis proves that utilisation of the time localisation properties of the Discrete Wavelet Transform is possible when taken together with a distance metric strategy. Appropriate results are presented to verify the performance of the Wavelet solutions provided by the framework, especially in respect of counteracting fraudulent coins in the recognition process. As an overall validation of the research solution, an emulation of the coin recognition system was produced that could validate coins in real time, this is also documented. Both the hardware and software components of the integrated framework which have been developed, are fully modular and hold significant potential for expansion and integration into newer, more powerful cost effective coin recognition systems.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:749628 |
| Date | January 1999 |
| Creators | Sharman, Darren |
| Publisher | University of South Wales |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://pure.southwales.ac.uk/en/studentthesis/investigation-and-application-of-digital-signal-processing-and-wavelet-technologies-to-automatic-coin-recognition(07e4df3d-fceb-4f0a-ad8a-411c1ac13ed9).html |
Page generated in 0.0013 seconds