Abstract
This work tackles the issue of the protection of confidential private data. Special attention is given to online banking. We made profit random functions to produce unpredictable results for intruders. By creating a list of passwords and by encrypting each word in two different steps based on random functions we obtain efficient double protection. Additional security is offered by biometrical data such as fingerprints. The fingerprints are watermarked in a well-chosen image which serves a shared key between the bank and its customer. This image is merged in the customer's fingerprints by using the mathematical principle of singular value decomposition (SVD). As a result we get a blurred image that is only understandable by the bank.