A biometric security is a technique by means of which digital contents are protected by acryptographic key generated from the biometric features of a person like Retina, Iris, Fingerprint,Face, Voice and so on. Normally the digital contents like documents are protected by acryptographic key generated from a unique password. The process in irreversible, i.e the key canbe generated from the password but not the vice versa. Passwords are relatively easy to hack asmost of the users keep their personal information like date of birth as password and alsopassword length has a limit as human beings cannot remember a password of significantly largelength. Hence guessing the password of a user, whose significant information is available, iseasier. Therefore off late lot of emphasis has been given to biometric features. Biometric featuresof no two people are same. For example the finger prints or the face of any two people differ.Hence if a template (alphanumeric or binary representation of features from a biometric data) isselected for the key generation than cracking them for accessing information becomessignificantly difficult. But as with every advantage comes certain limitations also. The keys are nottime invariant. Templates tends to change based on the data acquisition, or with time. Forexample the finger prints or palm prints changes with ages. Iris, retina and face features changeswith change in light intensity during the acquisition phase. Fingerprint features changes withchange in the orientation of the finger while scanning. In a classic authentication problem, suchvariability’s can be easily dealt with by keeping a threshold for the acceptance of the features.Such acceptance threshold is not applicable for the case of biometric templates. Even slightest ofthe variability in the templates changes the generated key, therefore causing a high falserejection rate. Hence in this work we analyze the most accepted biometric features andtechniques for key generation and propose the most invariable technique in terms of dataacquisition invariability. The work analyzes Iris, Face, Fingerprint and Palm prints for analysis ofthe biometric template generation and key generation form the templates. Further a uniquebenchmark analysis technique is proposed for quantifying the quality of a biometric model orfeatures.