Simple acoustical signature based coin validation

Abstract

The coin classification, recognition and validation is an important issue for vending machines and other coin handling equipment. One approach to investigate and classify the coins (often in combination with other methods-like optical and electromagnetic sensor signal processing) is the analyzing of the acoustical signature of the coin, falling against the special metal part (e.g. in the form of small plate or cylinder) of the coin validator, generating vibrations and sounds with special time-frequency domain properties. The characteristics like natural frequencies and their amplitudes of such interaction signals could be used for the classification and validation of the coins. It has found, that for the used mechanical setup the most appropriate is to use the frequency and the amplitude values of 1–3 maximum resonance peaks of the coin signature to classify and validate the coins. In the proposed solution values of the natural frequencies and the corresponding amplitudes are found by interpolation between the corresponding frequency bins of the relatively sparse and low sample-rate-based FFT, once per every interaction, allowing to have a simple and efficient solution, needing very small processing power. The proposed solution is described with comparison with other frequency-response-function based approaches with example plots for real coins. Also the required signal processing resources are estimated.