STM Journals Peer-reviewed research from Consortium e-Learning Network

Journal of Multimedia Technology & Recent Advancements

An Approach of Multi-modal Biometric Iris and Speech based Person Recognition System with Decision Fusion Technique

  1. Md. Rabiul Islam
  2. Md. Fayzur rahman

Abstract

This paper presents a unique approach of multi-modal iris and speech feature based person recognition system. Iris images and speech signals are taken from CASIA iris database and NOIZEUS speech database respectively. Iris features are extracted after applying iris images noise removing and image pre-processing techniques. On the other hand, speech signal noise removing, start-end points detection algorithm, silence parts removal, windowing and feature extraction techniques are applied to extract the features from the speech utterance. Feature fusion is performed by combining iris and speech based feature. Principal component analysis based dimension reduction technique has been used to uni-modal iris feature vector, uni-modal speech feature and iris-speech fused feature vector. Back-propagation neural network classification technique has been used to uni-modal iris, uni-modal speech and multi-modal iris-speech techniques. These three classification outputs are fed to the voting method based decision fusion approach. CASIA iris dataset and NOIZEUS speech dataset have been used to measure the performance of multi-modal iris-speech based decision fusion approach.Keywords: Iris-speech based multi-modal system, biometric technique, iris-speech based feature extraction, voting method  Cite this ArticleMd. Rabiul Islam, Md. Fayzur Rahman. An Approach of Multi-modal Biometric Iris and Speech based Person Recognition System with Decision Fusion Technique. Journal of Multimedia Technology & Recent Advancements. 2015; 2(2): 5–10p.
Support