Experimental Study of Sparse Watermarking Techniques for Multibiometric System

This paper focus on study and analysis of effect of watermarking technique in spatial domain combined with cs theory on verification and authentication performance of multibiometric system. In presented techniques, watermark fingerprint is compressed using CS theory before embedding into host biometric. These compressed fingerprint features embed into face image such that watermarked face image is used for verification and authentication of individual. Compressed fingerprint features are used for cross verification and authentication of individual. The modified LSB substitution based technique and modified correlation based technique using WGN combined with CS theory used to secure biometric data at system database and communication channel between two checkpoints of multibiometric system, respectively. The verification accuracy of multibiometric system using these watermarking techniques is around 96 % with more computational security and high perceptual quality of biometric data. The results show that these watermarking techniques do not have effect on authentication performance of multibiometric system. The novelty of paper is combined compressive sensing theory with watermarking technique for security of multibiometric data.