Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Development of a coin recognition application for Android Mobiles

Theodoroglaki Ioanna

Full record

---

Summary

This project aims at the classification of coins as true or counterfit. The key requirement for this thesis was that the recognition should be done efficiently, and, that the resulting system should be readily available to a broad range of different devices and platforms. After extensive research in the existing literature we have chosen the 2D Normalised Correlation or Cross-correlation algorithm to address the efficiency criterion. This algorithm is widely used for pattern recognition, electronic tomography, cryptanalysis and neuropsychology. In order to have a system which is readily available on many devices we chose the Android platform. This platform is an implementation of the Linux kernel for mobile devices (it is widely used in mobile phones, tablets, TVs and wearable devices such as smart watches). The system was fully developed and tested in a laboratory environment, with very good results. The present thesis briefly presents the 2D Normalised Correlation algorithm and the Android system. Subsequently, MATLAB models of the specific use of the algorithm for coin detection are presented, followed by the newly developed Android application. The hierarchy of menus is also presented in detail, as it matches the structure of the software. Performance results with an analysis of how various parameters of the application affect its performance, and conclusions from this work are presented as well.

Available Files

Services