A Secure Biometric Identification Technique Using Spread Spectrum Audio Watermarking

摘要

A biometric digital audio watermarking algorithmis presented in this paper. Instead of a constant watermark, a biometric watermark that is unique to an individual isembedded into an audio. Keystroke Dynamics is chosen to stampthe ownership of an individual to the audio file. Constructedbiometric template is embedded into wavelet domain by usingfrequency hopping spread spectrum technique. Direct SequenceSpread Spectrum technique is implemented before embeddingprocess. The extraction of the biometric template is a blindprocess, in other words, original audio carrier is not requiredto extract the watermark. The experimental results illustratedthat the system is robust against the signal processing operationssuch as low-pass filter, adding white Gaussian noise, shearing, and compression. Furthermore, the audibility of this techniqueis in plausible values. After extraction of the biometric data, a biometric identification system is constructed. The correctidentification rates were presented.