Sheltered identification with hiding biometrics

Biometrics based personal identification techniques have intrinsic advantages over traditional personal identification techniques but the security of biometric data is a major issue. The biometric based verification system works properly only if the verification system can guarantee that the biometric data has come from the genuine person at the time of enrollment. We introduced application in which we hide biometric feature of user in variety of images in order to enhance security of biometric template. We have proposed an image adaptive hiding method which adjusts the strength of the biometric bits based on host image property. Due to the redundancy of watermark bit, we are able to retrieve the hidden data with 100 percent accuracy with limited payload. Biometric watermark used for this work is fingerprint minutiae. However the idea can be extended for any other feature or combination of more than one feature. We have used a distortion-tolerant matching algorithm that defines a novel feature vector for each fingerprint minutia based on the global orientation field, similar to the method proposed by Jin Qi and Yangsheng Wang. These features are used to identify corresponding minutiae between two fingerprint impressions by computing the Euclidean distance between vectors that gives high verification accuracy. The algorithm proposed here is tested for 50 different types of host images and public data collection, DB3, FVC2002.