Abstract

Replay attack on biometric pose a challenge to the security of biometric recognition systems. Many\nresearchers have proposed the algorithms for watermarking the biometric data to mitigate replay attacks but most of\nthem usually compromise with the recognition performance of biometrics. In this paper, an algorithm that selects an\noptimal bit plane using cuckoo based optimization is proposed, to hide the watermark in iris biometric using reversible\ninteger wavelet transform. The algorithm minimizes the compression ratio of image, leading to invisible watermarking.\nThe use of reversible watermarking has an advantage of retrieving the original biometric back from the watermarked\nimage without any loss of precision and impact on recognition performance. The proposed algorithm uses the sensor ID\ncombined with current date and time as watermark. This choice of watermark propels the verification at the extraction\nend and hence detects whether it is replayed data or the fresh sample. The recognition performance of iris before and\nafter watermarking has also been compared and the results are found to be same.