Sorting and counterfeit detection of philippine banknotes for vending machines using surf feature extraction and graphbased pattern recognition/ Mary Jane P. Calulang, Jeanne May H. Carolino, Maria Evita M. Juan, John Paul S. Monter, and Vincent Johanne P. Tenorio.--

Bachelor's thesis

College of Science.--
Bachelor of science in computer science: Technological University of the Philippines,
2025.

Includes bibliographic references and index.

This study covers the development and testing of a system created to sort and
authenticate Philippine banknotes meant for vending machines. Using SURF features for
finding image features and graph-based pattern identification for classification, the
system is designed to cope with the growing trend of secure and efficient verification of
currency at unattended stores. It carried out a series of planned experiments to check the
system’s performance for both new and old banknotes. Detecting banknotes resulted in a
91.67% accuracy, having greater precision than recall except for the most valuable notes.
In addition to assessing the system, a balanced group of technical and non-technical
respondents shared what they thought of its usability, reliability and chances for use in
the real world. Results from tests with consumers reveal that most people are very
positive about how the machines function, how innovative they are and how they fit into
existing systems. Research shows that new banknote styles are easier for people to
identify, matching the characterization criteria the algorithms require. This research
demonstrates that using both SURF and graph techniques in financial systems works well
and shows where more improvements can be made in terms of solidity and future
performance.