| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 405857 | Neurocomputing | 2016 | 13 Pages |
•A Joint Encryption and Compression (JEC) technique for transmission of biometric data over a wireless channel is proposed.•An algorithm is suggested for fusion of individual biometric scores in order to improve the overall recognition performance.•The JEC scheme is tested for its suitability in a wireless communication system. It can be seen that with the help of appropriate channel coding high quality of reconstruction (PSNR) can be ensured.•The recovered biometric data offers recognition rates that are found to be acceptable for a telebiometric system.•It is concluded that JEC in combination with biometric fusion can lead to an effective multimodal telebiometric system.
In this paper we present a novel Joint Encryption and Compression (JEC) technique for transmission of biometric data over a wireless channel. The method offers advantages such as reduced data processing, security and enhanced recognition accuracy. The security of the biometric data is ensured by means of watermarking followed by random bit shuffling. The watermarking process involves embedding one׳s fingerprint information in his/her compressed face image. In order to compress face images face images wavelet based encoders such as SPIHT (Set Partitioning Hierarchical Trees) and JPEG2000 are used. The advantage of the proposed method is that the overall data rate can be minimized while simultaneously maintaining good quality reconstruction. The JEC (Joint Encryption and Compression) scheme is tested for its suitability in a wireless communication system. It can be seen that with the help of appropriate channel coding, high values of PSNR (Peak Signal to Noise Ratio) can be obtained. Consequently the recovered biometric data offers recognition rates that are found to be acceptable for a telebiometric system. Additionally an algorithm is suggested for fusion of individual biometric scores in order to improve the overall recognition performance. Simulation results depict that a significant advantage in verification rates can be achieved by this method.
