Online fortune-telling system using angle difference

• Main Author: Tan, Zhong Ming.
• Other Authors: Li Fang
• Format: Final Year Project
• Language: English
• Published: 2010
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Pattern recognition
• Online Access: http://hdl.handle.net/10356/39760
• Institution: Nanyang Technological University

In this project, the biggest concern is that the time-efficiency of the matching algorithm suffers considerably when handling large data sets. We have proposed various indexing algorithms to help reduce the computational cost incurred. From the experiment results, discrete Fourier transform turned out to incur the most computational cost, therefore we used DFT as a standard in our comparisons. Out of the various indexing algorithms, FFTW[13] (a free library provided on the internet, widely used by the community) provided an approximately 577% reduction in computational cost as compared to the discrete Fourier transform that was previously employed. Following next, fast Fourier transform provided the 2nd highest amount of computational cost reduction of approximately 568% as compared to the discrete Fourier transform. This reduction in computational cost came without any reduction in the uniqueness of the output data. Recommendations were provided, as both FFTW and FFT are the choices that users can select from. Depending on the nature of the application, both have pros and cons but at the same time maintaining similar computational cost reduction capability. The objectives of the project were well achieved. Keywords: palm print, biometric identification, indexing algorithms, time-efficiency, fast Fourier transform, DFT(discrete Fourier transform), FFTW[13], computational cost.