Multimodal biometrics recognition based on image latent semantic analysis and extreme learning machine

Abstract

Multimodal biometrics recognition system suffers from the shortcomings of large data processing and much time cost during the recognition. To overcome the shortcomings of the traditional methods, in this paper, a novel multimodal biometrics recognition method is proposed by using image latent semantic analysis and extreme learning machine method. The image latent semantic analysis for multimodal biometrics feature extraction will extract abandon information from the images and the extreme learning machine method has the merits of high accuracy and fast speed. With this new method, the latent semantic features from the multimodal biometrics images are digged out to improve the recognition accuracy. Finally, the extreme learning machine is used as the classifier. The experiments show that the proposed algorithm has get better performances both in recognition accuracy and speed. © Springer International Publishing 2013.