Authors: Nirmala Saini; Aloka Sinha
Addresses: Department of Physics, Indian Institute of Technology Delhi, Hauz Khas, New Delhi – 110016, India ' Department of Physics, Indian Institute of Technology Delhi, Hauz Khas, New Delhi – 110016, India
Abstract: In this paper, a new transform Gabor filtered Wigner transform (GFWT) has been proposed. In GFWT, Gabor filtering is performed on the Wigner transformed image. Wigner transform gives a simultaneous representation of an image in time and frequency domain which is further processed using Gabor filters. The proposed transform is then used to extract the features from the biometrics to develop different multimodal biometric systems. A detailed study has been carried out in which, different unimodal and multimodal systems such as feature level and score level fusion are analysed. In order to improve the performance of the system, an optimisation technique, particle swarm optimisation (PSO) is used to find the optimal parameters of the Gabor filter and to select the significant GFWT feature vector. The PSO technique not only improves the performance of the system but also able to reduce the dimension of the feature vectors. Numerical experiments are carried out on face and palmprint database to show the effectiveness of the proposed transform for different unimodal and multimodal systems.
Keywords: multimodal system; feature level fusion; score level fusion; Gabor filtered Wigner transform; particle swarm optimisation; PSO.
International Journal of Biometrics, 2020 Vol.12 No.3, pp.301 - 316
Accepted: 14 Dec 2019
Published online: 24 Jun 2020 *