Summary
International Symposium on Nonlinear Theory and its Applications
2008
Session Number:A2L-D
Session:
Number:A2L-D1
Simplified DFT for Hand Posture Recognition System
Hiroomi Hikawa, Hirotada Fujimura,
pp.-
Publication Date:2008/9/7
Online ISSN:2188-5079
DOI:10.34385/proc.42.A2L-D1
PDF download (172.5KB)
Summary:
This paper describes simple discrete Fourier transforms (DFT) to be used in the hand posture recognition system. Conventional DFT uses sine cosine, and square root functions, which requires high computational and hardware costs. The new DFT uses heavily quantized version of sine cosine functions that take only three values, resulting the generation of the function becomes easier and multiply add (MA) operation is replaced by simple add-sub-zero operation. For the calculation of the magnitude value, absolute and add operations are used instead of the square root function. Due to the simple function, the recognition performance is degraded, but the simulation results shows that the deterioration of the recognition rate can be suppressed to about 1% by using training data set with the images in different positions.